Interactions of a Light Hypersonic Jet with a Non-Uniform Interstellar Medium
Ralph S.Sutherland and Geo?rey V.Bicknell
Research School of Astronomy and Astrophysics,
Australian National University,ACT 0200,Australia.
ABSTRACT
We present three dimensional simulations of the interaction of a light hypersonic jet with an inhomogeneous thermal and turbulently supported disk in an elliptical galaxy.These simulations are applicable to the GPS/CSS phase of extragalactic radio sources.The interstellar medium in these simulations consists of a conventional hot (T ～107K)component together with a warm (T ～104K )turbulently supported disk whose local density is described by a log-normal density distribution and whose spatial structure is realized from a fractal power-law.We model the jet as a light,supersonic non-relativistic ?ow with parameters selected to be consistent with a relativistic jet with kinetic power just above the FR1/FR2break.We identify four generic phases in the evolution of such a jet with the inhomogeneous in-terstellar medium:1)an initial “?ood and channel”phase,where progress is characterized by high pressure gas ?nding changing weak points in the ISM,?owing through channels that form and re-form over time,2)a spherical,energy-driven bubble phase,were the bubble is larger than the disk scale,but the jet remains fully disrupted close to the nucleus,3)a subsequent,rapid,jet break–out phase the jet breaks free of the last obstructing dense clouds,becomes col-limated once more and pierces the spherical bubble,and 4)a classical phase,the jet propagates in a momentum-dominated fashion similar to jets in single component hot haloes,leading to the classical jet –cocoon –bow-shock structure.Mass transport in the simulations is investigated,and we propose a model for the morphology and component proper motions in the well-studied Compact Symmetric Object 4C31.04.Subject headings:Radio Galaxies —GPS/CSS,ISM —Turbulence,Fractal medium.1.Introduction There is a substantial literature on jet simulations going back to the time when the ?rst supercomputers
became available for computational astrophysics (Norman et al.1982).The simulations of extragalactic radio jets that have been carried out typically have axisymmetric or cartesian slab-jet geometries;a smaller number of three dimensional simulations have also been conducted.One of the most important features has generally been the assumption of a hot,tenuous and smoothly distributed ambient atmosphere.These simulations have been extremely informative,leading to signi?cant insights into the physics of extragalactic jets.For example the production of jet knots and ?lamentary structure has been associated with non-linear development of Kelvin-Helmholtz modes (Norman et al.1984;Hardee et al.1992);slab-jet simulations have shown how the extragalactic dentist drill (Scheuer 1982)may work in practice (Hardee &Norman 1990)and simulations of jets in atmospheres with density gradients (Hardee et al.1991)have shown how the stability/instability of jets is in?uenced by the gradients in the ambient atmosphere.More recently,Krause (2003)has mapped out a region of the parameter space of jet density ratio,ηand Mach number,M and has
a r X i v :0707.3668v 1 [a s t r o -p h ] 25 J u l 2007
also simulated the progress of a light jet through a dense(but uniform)medium typical of Gigahertz Peak Spectrum(GPS)or Compact Steep Spectrum(CSS)radio sources.
Notwithstanding the insights provided by the above work,until recently one key element has not been addressed in detail and this is the nature of the background medium.In the case of classical double radio sources,there is ample justi?cation for assuming an homogeneous medium.However,there is excellent observational motivation for the disruption or modi?cation of radio source morphologies by an inhomogeneous medium.For example,the existence of?lamentary optical line-emitting gas adjacent to the inner radio lobes in M87has been known for some time(Ford&Butcher1979)as has the hierarchical structure of M87 evident in both radio and X-ray images(Owen et al.2000;B¨o hringer et al.1995).Moreover,recent Chandra observations of the inner few kpc of M87,indicate signi?cant interaction of the inner radio plasma in M87 with the hot X-ray gas leading to the production of X-ray?laments.
There is also evidence for jet–ISM interaction in high redshift radio galaxies such as4C41.17Bicknell et al.(2000),quasars such as3C48(Wilkinson et al.1991),blazars such as MKN501(Giroletti et al.2004; Bicknell et al.2005),and the general class of GPS and CSS sources(De Vries et al.1999;Bicknell et al. 1997),which motivated our current line of research in the?rst place.All of these sources show clear evidence for the interaction of a jet with a clumpy medium which substantially distorts the morphology of the radio source,in most cases producing accompanying emission line luminosity from the disturbed dense gas.
There are other compelling reasons for considering the interaction of radio jets with an inhomogeneous medium.Ever since the paper of Silk&Rees(1998),proposing an explanation for the Magorrian et al. (1998)relation between black hole mass and bulge mass,there has been an increasing appreciation of the importance of feedback processes involving active galactic nuclei and the evolving interstellar medium of forming galaxies.In principle,jets,winds and radiation emitted from the environs of the black hole can impede continued accretion into a forming galaxy thereby limiting its growth.The importance of jets stems from the realisation that disrupted jets can process4πsteradians of solid angle in the interstellar medium (e.g.Saxton et al.(2005))and that they transport a large amount of momentum.For similar reasons the importance of jet–ISM interactions has been appreciated in attempts to resolve the questions posed by cooling?ows as well as the X-ray cavities associated with radio bubbles that are now being observed in these environments(e.g.Nulsen et al.(2005)).
In recent work we have begun to investigate these phenomena through the simulation of two-dimensional slab jets issuing into an inhomogeneous medium(Bicknell et al.2003,?;Saxton et al.2005).We have shown that a numerous di?erent radio morphologies can arise as a result of the interaction of both the jet and lobes with a clumpy interstellar medium and we have compared the resulting radio morphologies with a number of well–known radio sources.Some preliminary three-dimensional simulations have also been published (Sutherland2005;Bicknell2006).
In this paper we present a high resolution simulation of a radio jet interacting with an inhomogeneous interstellar medium in the form of a turbulently supported disk which extends our previous work in several di?erent directions:(1)The simulation is three dimensional with a resolution of512cells in the(cartesian) coordinate directions and a spatial scale of2parsecs per cell.(2)The density structure of the warm disk is described by a log-normal distribution,typical of the warm interstellar medium in a number of di?erent environments.(3)The initial distribution of the warm medium is that of a thick,almost Keplerian disk supported by a combination of thermal pressure and a,dominant,supersonic velocity dispersion.This type of distribution is one of several that may be contemplated and is supported by the observation of such disks in M87(Dopita et al.1997)and NGC7052(van der Marel&van den Bosch1998).(Other types of
distributions of warm gas that may be contemplated include that typical of the Lyman-αhaloes observed in many high redshift radio galaxies.These will be considered in future papers in this series.)(4)The hard and soft X-ray emissivity and surface brightness of the thermal plasma is calculated,in conjunction with the radio surface brightness image.These synthetic images provide valuable insights for the interpretation of observational data.
This simulation illustrates four important phases in the evolution of a young radio galaxy:(1)An initial “?ood and channel”phase wherein the radio source is beginning to be established,but is still interacting strongly with the disk material.(2)A quasi-spherical jet-driven bubble phase where the jet is fully disrupted and drives a pseudo-spherical bubble into the surrounding medium.(3)a rapid jet establishment phase where the last obstruction is ablated away and the jet reforms and crosses the spherical bubble rabidly until it breaks out.(4)A classical jet–cocoon–bow-shock phase familiar from previous studies of radio sources in a single component,hot ISM.We also determine the X-ray morphology associated with the interaction of the radio source with the disk and show that the luminosity from the disk may persist.
We begin,in the following section,by describing the key elements and assumption in some detail.We discuss our justi?cation of,and constraints on,the parameters chosen for the simulations.
2.The Jet-Galaxy System:Key Model Element and Parameters
2.1.Host galaxy potential
Elliptical galaxies contain both baryonic and dark matter.The former dominates on scales close to the core and the latter dominates at large radii,say of order10kpc.Since jet simulations frequently extend over this range of scales,we have constructed a family of potentials which we use to represent a combined, self-consistent distribution of baryonic(luminous)and dark matter.In the simulation presented here,the scale of the simulation is such that the potential is dominated by the baryonic component close to the core. However,the dark matter does have some in?uence.Hence,we present here a generic potential that is useful here and which can also be used in future simulations with a larger overall spatial scale,in which the dark matter has an even greater in?uence.The potential is based on isothermal distributions of baryonic and dark matter,described in terms of isotropic distribution functions,as follows.
Let,(σB,ρB,0)and(σD,ρD,0)be the line of sight velocity dispersions and central densities of baryonic and dark matter respectively and let the total speci?c energy of a baryonic particle(star,gas),or dark matter ‘particle’,be E=1/2v2+φG where v is the velocity andφG is the gravitational potential.The central value of the potential is taken as,φG(0)=0.We assume Maxwellian distributions for the distribution functions, for both dark and baryonic matter:
f D(E)=
ρB,0
(2π)3/2σ3
B
exp(?E/σ2D),
f B(E)=
ρD,0
(2π)3/2σ3
D
exp(?E/σ2B).(1)
As for the classic single component isothermal sphere(c.f.King(1966)),we de?ne dark and baryonic matter core radii,r D and r B,via the relations:
4πGρD,0r2
D
σ2
D =
4πGρB,0r2
B
σ2
B
=9.(2)
Forκ=σD/σB 1andλ=r D/r B 1,the density is dominated by the baryonic component near r=0 and by the dark component near r=r D.In this case,the core radii have the conventional interpretation of being the radii at which the surface densities of the respective components drop to approximately half of their central values.
Taking =ρB,0/ρD,0,the dark and baryonic matter densities are given in terms of the dimensionless potentialψ=φ/σ2
D
by:
ρD
ρD,0
=exp(?ψ),
ρB B,0=
ρB,0
D,0
exp
?
σD
B
2
ψ
(3) = exp
?κ2ψ
.(4)
De?ning the normalized radius by r =r/r D,the dimensionless version of Poisson’s equation is:
d2ψdr 2+
2
r
dψ
dr
=9
exp(?ψ)+ exp
?κ2ψ
.(5)
We may take the parameters of this double isothermal distribution to beκ,the ratio of the dark to baryonic velocity dispersions,and ,the central baryonic to dark matter densities.However,it is generally more convenient to takeκ,and the ratio of core radii,λ,as de?ning parameters,use =λ2/κ2,and?nd the central densities from equations(2),
ρD,0=
9σ2
D
4πGr2
D
,ρB,0=
9σ2
B
4πGr2
B
.(6)
We refer below to the double potential function for givenκandλparameters asψκ,λ.
Equation(5)is numerically integrated for chosen parameters,tabulated and interpolated with cubic splines when required by the hydrodynamic code.In the present modeling we useκ=2,andλ=10, referring to the potential asψ2,10.The central density is dominated by baryonic material over dark matter by a ratio of25:1.Figure1in the next section shows the resulting equilibrium density distribution of the hot galactic atmosphere in theψ2,10potential,compared to two single isothermal potentials.
2.2.The Hot Galactic Atmosphere
Generally,in radio galaxies,there are at least two components of the interstellar medium,a tenuous, smoothly distributed hot atmosphere with central density～0.01?1particles cm?3and T～107K,and a cold/warm unevenly distributed component with number densities～1?100cm?3and temperatures ～103?4K.
The tenuous,hot interstellar medium in our simulations is isothermal,although it is relatively straight-forward to relax this condition by adopting,for example,an empirically determined radial temperature pro?le.Let T be the gas temperature.The virial temperature corresponding to the dark matter is de?ned by T?=ˉmσ2
D
/k,whereˉm=μm u≈1.035×10?24g,is the mean particle mass(including electrons,hydrogen and heavier ions),usingμ=0.6224,the mean molecular weight in fully ionised solar metallicity gas,and
m u,the atomic mass unit.Assuming hydrostatic equilibrium,the particle number density,n h of the hot gas is given simply by:
n h n h,0=exp
?
T?
T
ψ
.(7)
(Note that this de?nes the total particle density,not the Hydrogen number density,often denoted by n H). On the1kpc scales simulated here,the sound crossing time for the region at400kms?1is about2.5Myr,so the hydrostatic formalism used requires the atmosphere to have settled into place for longer than this prior to the jet interaction,which we assume to be the case.
Figure1shows how the T=T?hot atmosphere?lls the double isothermal,ψ2,10,potential with r D=3.5kpc.For comparison,corresponding functions for single isothermal potentials,one each for a core radius of r c=3.5kpc,and r c=350pc,are overplotted.The upper logarithmic plot shows the large scale behavior,and the lower linear panel focuses on the1kpc range used in the simulations here.The key feature is that between the inner radius and outer radius,the density decreases with radius with an intermediate slope.Within the1kpc range on the grid,the hot atmosphere density has fallen to approximately40%of its central value,and is essentially uniform(within10%)inside200pc,and is described by a radial powerlaw with a slope of index?0.68in the outer regions to1.0kpc.
2.3.The non-uniform,fractal,warm interstellar medium
The warm and cold interstellar medium has been known to be non-uniform ever since the?rst obser-vations of nebulae in the late19th and early20th centuries(although recognition of an interstellar medium as such came later in the20th century).In these simulations we introduce a non-uniform medium,which is describable semi-analytically,with the view to assessing qualitatively the in?uence on inhomogeneity on the energetics of dynamical interactions–even if a full theoretical understanding of the non-uniformity un-available.We compare a homogeneous model with an inhomogeneous model,at two resolutions,in order to comprehend the consequences of non-uniformity will take,in a global dynamical sense.
To establish non-uniform medium we make use of an obvious analogy with a turbulent medium.The physics literature on turbulence is vast,with results and models in a many?elds of physical sciences,and astrophysics is is no exception.Rather than attempt to review this huge topic here however,we refer the reader to the recent astrophysically oriented annual reviews of(Elmegreen&Scalo2004),and(Scalo& Elmegreen2004)as a starting point for background material,and then refer simply to some speci?c results that we use below to arrive at reasonable(although not necessarily unique)parameters for our non-uniform medium.
It must be emphasized that for the present work we are not modelling actual turbulence in a genuine causally generated ISM(see Kritsuk et al.(2006)for a recent large scale isothermal ISM turbulence simula-tion).Rather,we are taking a parameterisation for the non-uniform properties of generic turbulent media focusing on characteristics such as the variance,σ2,the intermittency(in skewed distributions),and two–point self-similar power-law structures,relying on a range of previous experimental and theoretical results from the?eld of turbulence.Hence the initial distribution of the ISM that we employ should be regarded as a physically motivated generalisation of a homogeneous model,whilst not necessarily representing an accurate physical model of a turbulent ISM.
Fig.1.—The density distribution of the normalized two component isothermal potential,ψ2,10.The solid curve represents the normalized total density distributionρ/ρ0=exp[ψ2,10],scaled to a dark matter radius of r D=3.5kpc,implying a baryonic core radius of r B=350pc.The short dashed curve is the density distribution corresponding to an isothermal potential with a core radius of3.5kpc.The dash-dot curve is a density distribution for an isothermal sphere with a core radius of350pc.The upper panel uses logarithmic coordinates,the lower panel is linear and covers the domain of the simulated hot atmosphere.See
text for details.
2.3.1.Log-normal density distribution
Turbulence naturally gives rise to non-uniform structure,in velocity and density?elds.We neglect the velocity structure,and focus on choosing statistical parameters to describe the density.This is justi?ed numerically by the relatively small turbulent velocities expected when compared to the very large velocities found in our global jet–ISM interaction.The velocities observed in typical warm ISM conditions fall in a range of transonic to mildly supersonic values,Mach1-4(e.g.Heiles(2004)).At temperatures at or below104K this corresponds to velocities<50km s?1.In our1kpc simulations over105yr timescales, the resulting displacements amount to less than2or3cells,and are insigni?cant compared to those of the radio jet and the energy bubble/cocoon generated by the main outburst,wherein velocities of hundreds or thousands of km/s occur.Consequently,we do not impose a turbulent velocity?eld on the warm medium, and focus instead on constructing the density?eld.
We use a log–normal distribution to describe the single point statistics of the density?eld of our nonuniform ISM.The log–normal distribution is a skewed continuous probability distribution.Unlike the normal distribution,it has a non-zero skewness,variable kurtosis,and in general the mode,median and mean are unequal.The log-normal distribution appears to be a nearly universal property of isothermal turbulent media in experimental,numerical and analytical studies(e.g.Nordlund&Padoan(1999),see also Warhaft (2000),and Pumir(1994)).Moreover,it is encouraging that the log–normal distribution is the limiting distribution for the product of random increments,in the same way that the normal distribution plays that role for additive random increments.It is thus compatible,at least conceptually,with a generic cascading process consisting of repeated folding and stretching.
We begin by describing parameters for an inhomogeneous interstellar medium density?eld,which is on average isotropic.That is,there is no dependence on location.In§2.3.4below,we describe how this standard distribution is modi?ed to re?ect the potential.With a log–normal distribution,the natural logarithm of the ISM density?eld is a Gaussian which has a mean m and variance s2.The probability density function for the log-normal distribution of the mass densityρis,
P(ρ)=
1
s
√
2πρ
exp
?(lnρ?m)2
.(8)
The meanμand varianceσ2of the density are given by
μ=exp[m+s2/2].(9)
σ2=μ2(exp[s2]?1)(10)
Additional statistical properties of the log-normal distribution are summarized in Appendix A.
In this simulation we adoptμ=1.0,σ2=5.0,as our standard log–normal distribution.These values are compatible with the favored ranges in Fischera et al.(2003)and Fischera&Dopita(2004)from star burst galaxy reddening and extinction considerations.The variance measures how concentrated the mass is in dense cores,or conversely how much volume is occupied by voids,and these parameters give a?atness parameter F=1836,indicative of an intermittent distribution(see Appendix A).With these parameters, densities below the mean,z=μ,comprise one quarter of the mass,and occupy three quarters of the volume, and the mean is approximately20times the mode(see§C for further details).Other values are possible of course,and a proposed relationship between the log-normal variance and the isothermal Mach number of the turbulence given by(Nordlund&Padoan1999),viz.
σ2≈0.25M2,(11)
suggests that higher mach numbers in an AGN medium for example,could be compatible with larger variances.Below,when we use turbulent velocity support to determine the scale height of the warm ISM disk in the galactic potential,we adopt a turbulent velocity that is consistent with the valueσ2=5.0adopted here;hence the turbulent velocity is not a completely independent parameter in the model.
2.3.2.Power-law density structure
The two–point structure of a homogeneous turbulent medium is best described in Fourier space.We denote the Fourier transform of the densityρ(r)by F(k)(where k is the wavenumber vector).The isotropic power spectrum D(k)is the integral over solid angle in Fourier space of the spectral density F(k)F?(k).In
three dimensions:
D(k)=
k2F(k)F?(k)d?.(12)
.Even if the spectral density is anisotropic,the angular integral averages the spectral density into a one dimensional function of k only.
For a power-law dependence on k,D(k)∝k?βandβ=5/3,the spectrum is referred to as Kolmogorov turbulence.It has been shown that a scalar tracer(density)of the turbulent?eld also shows the Kolmogorov structure index(Warhaft2000).
We follow Fischera et al.(2003)and adopt a standard density power spectrum with a Kolmogorov power-law withβ=5/3to generate a spatial structure power-law for the density in our non-uniform ISM. Given our assumption above that the(isothermal)disk is mainly supported by supersonic turbulence,a slightly di?erent value ofβ=2.0associated with shock turbulence may be a reasonable alternative(e.g. Boldyrev et al.(2004))but we postpone investigation of the variation ofβto future work.
2.3.3.Iterative Generation and Fractal Resolution
For our inhomogeneous ISM we selectμ=1.0andσ2=5.0as our standard log–normal density?eld parameters,and a Kolmogorovβ=5/3power-law spatial index.This structure is modi?ed as described below in§2.3.4to take account of a spatially variable mean density.The selected parameters may be representative in view of the results of Fischera et al.(2003)and Fischera&Dopita(2004)noted above. However,at this stage not a lot is known about the parameters of the inhomogeneous ISM of radio galaxies.
In order to simultaneously achieve log-normal single-point statistics and a power-law self-similar struc-ture,we have implemented the practical method developed by atmospheric scientists,for constructing two and three dimensional terrestrial cloud models,which are used in radiative transfer calculations(Lewis& Austin2002).The?rst step of the procedure involves constructing a cube in which each cell has a Gaussian distribution of mean zero.This cube is Fourier transformed and then apodized by a power-law in wave number.The apodized cube is then Fourier transformed back to the spatial domain and retains its Gaussian statistics because a sum of Gaussians is still a Gaussian.A cube with log-normal statisitics is then produced by exponentiating the Gaussian cube.However,as re-transformation to Fourier space shows this alters the power-law in wave number property.Essentially the generation of dense cores and large voids consistent with the log-normal parameters,alters the low and high wavenumber ends of an original power-law,breaking self-similarity.Hence,the deviation from a power-law is calculated and this is used to estimate a correction to the power-law of the Gaussian distribution.Successive corrections are applied until satisfactory convergence
to a power-law(within approximately1%)is obtained,usually in about4–6iterations for distributions with modest variances(σ2<10).The process converges more slowly as the target variance,σ2,increases,and variances>10were not attempted.The reader is referred to Lewis&Austin(2002)for further details of this method.
From this process a library of cubes with a range of resolutions and variances were pre-computed for use in a range of simulations,from which our standard set:μ=1.0andσ2=5.0,β=5/3was selected.
The remaining choice in this procedure is to select the range of wave numbers over which to generate the fractal,in particular the minimum wave number k min,which determines the largest structure scale in the resulting fractal with respect to the spatial grid.With the scale height of the disk in the simulations being of order100pc in a domain of1kpc extent,a minimum wave number of k min=20is used to ensure that the largest‘clouds’are of order50pc,and appropriate to the scale of the disk everywhere.In the?nal structure used to form the disk,the wave numbers in the Fourier domain range from k=20to the Nyquist limit, N/2?1=255,covering just over1.1decades of scale.This is limited by the computational requirements for the overall grid,and the fact that the disk is only a small part of the domain.Signi?cant improvement requires a substantial increase in computing resources than are available to treat more than say2decades of structure in the fractal disk alone.
That said however,we did perform the simulation with two resolutions,low and high,discussed in§3.3, to investigate resolution dependent di?erences,and the di?erences of the low and high resolution fractal simulations from a uniform model.The evolution of these di?erent simulations is evidence that at least some of the fractal properties are being captured.
2.3.4.Equilibrium turbulent disks
As noted in§1many radio galaxies exhibit a turbulent disk of gas in the central regions,motivating us to consider the interaction of a jet with such a disk.Let us now consider the establishment of a disk-like distribution of gas in some detail.
There are two elements to establishing a turbulent disk.We begin with the fractal,power-law distribu-tion described above.This may be scaled to give a fractal distribution with a speci?c,but constant,mean density.This is unsatisfactory since the distribution of gas would not re?ect the potential of the galaxy. Hence,we derive below expressions for the mean density of a turbulent gas disk in the potential discussed in§2.1.We use the spatially dependent mean density to scale the fractal cube.Essentially this provides a single realization from an ensemble of turbulent fractal disks.
Let the density of the warm disk gas beρand its velocity be v i.We express the density and velocity as statistical averages,as follows,with the angle brackets expressing ensemble averages:
ρ=ˉρ+ρ ρ =0,
(13)
v i=?v i+v i ρv i =0.
For a recent description of this mass-averaged approach to a statistical description of turbulent?ow,see Kuncic&Bicknell(2004).We derive the relevant equations for our warm clumpy disk in the appendix, showing en-route that the azimuthal velocity in the turbulent disk is a function of the cylindrical radius only.
Following,Strickland&Stevens(2000)we also adopt the ansatz of an almost Keplerian disk.Let
v K=(r?φ(r,0)/?r)1/2be the Keplerian velocity in the disk mid-plane and put
?vφ=e K v K=e KσD
r?ψ(r,0)
?r
1/2
,(14)
where e K is a constant close to unity.
Taking the mean temperature of the disk to be?T and the line-of-sight turbulent velocity dispersion to beσt,the mean density of gas in the potential is given by:
ˉρ(r,z)ˉρ(0,0)=exp
?
σ2
D
σ2g
ψ(r,z)?e2ψ(r,0)?(1?e2)ψ(0,0)
.(15)
whereσ2g=σ2t+k?T/μm and as previouslyσD is the velocity dispersion of the dark matter.(Equation(15) assumes that?T andσt are constant throughout the disk.)
The main di?erences from the development of a similar equation by Strickland&Stevens(2000)is the (mandatory)lack of dependence of e K on z and the formal introduction of a turbulent velocity.The latter avoids the di?culty of prescribing an unphysically large temperature in order to achieve a reasonable disk scale height.
In summary then,we use the double isothermal potentialψκ,λwithκ=2,λ=10,scaled by r D=3.5kpc, andσD=400km/s,and a rotation parameter e K=0.93,plus a velocity dispersion of the warm gas of σg=40km/s for the disk.This velocity dispersion corresponds to an adiabatic Mach number,M～4, which is approximately consistent with the log–normal variance parameterσ2=5.0we adopted for the density?eld,if the Nordlund&Padoan(1999)relation(equation(11)holds.
3.Jet Parameters
We use a non-relativistic code for these simulations.Other aspects of the code are discussed in§4.The use of such a code to simulate phenomena that involve relativistic?ow in parts of the grid,is not ideal. Nevertheless,a large part of the?ow?eld is non-relativistic and is driven by the energy or momentum ?ux provided by the jet.Hence,we establish jet parameters in such a way that the energy?ux of the non-relativistic jet corresponds to the jet energy?ux of a given relativistic jet using relationships derived by Komissarov&Falle(1996).
The relationships between the respective densities and Mach numbers are given by the following rela-tionships for relativistic and non-relativistic jets with the same energy?ux,velocity and pressure:
ρnr=
2γ
γ?1
p
c2
Γ2
1+
Γ
Γ+1
χ
.(16)
M2nr=2M2
rel
2?γ
1+
Γ
Γ+1
χ
1+
χ
2?γ
?1
.(17)
(Komissarov&Falle1996).We use these relationships to determine the ratio of the non-relativistic jet density to the background and the non-relativistic jet Mach number.First note that the relativistic Mach number is given by:
M2rel=2?γ
γ?1
Γ2β2
1+
χ
2?γ
.(18)
As a result of the above relationships,the following equations for the conventional non-relativistic jet pa-rameters,the density ratio,ηand the Mach number,M nr are derived.The parametersξand T ism appearing in equation(19)are the ratio of jet to external pressures and the(hot)interstellar medium temperature respectively.
η=
2γ
γ?1
ξ
kT ISM
μmc2
Γ2
1+
Γ
Γ+1
χ
.(19)
M2nr=
2
γ?1
1+
Γ
Γ+1
χ
Γ2?1
.(20)
Note that the low value of kT/ˉmc2～10?6guarantees a light non-relativistic jet(η 1)despite a high Lorentz factor,and that the non-relativistic Mach number e?ectively corresponds to Lorentz factor.
In our simulations we take the gas to have the ideal adiabatic index,γ=5/3,since this represents the external medium the most accurately and we are mainly interested in the e?ect that the jets have on the external clumpy medium.
The following expressions for the(equivalent)versions of the jet power are also useful.Let D represent the diameter of the jet with cross-sectional area A=πD2/4.Then,the relativistic and non-relativistic jet powers are given by:
F E,rel=
γ
γ?1
cp jet AΓ2β
1+
Γ?1
Γ
χ
,
=3.9×1040
γ
γ?1
ξ
p ism/k
107
D
10pc
2
Γ2β
1+
Γ?1
Γ
χ
ergs s?1.(21)
F E,nr=
γ
γ?1
p jet vA
1+
γ?1
2
M2nr
,
=3.9×1040γ?1
ξ
p ism/k
D
2
β
1+
γ?1
M2nr
ergs s?1.(22)
A point to note from these equations,which is relevant to the choice of jet parameters,is that for a given jet Lorentz factor and the ratioχof rest-energy density to enthalpy,the jet power is proportional to the ratioξof jet pressure to ISM pressure times the ISM pressure.Even for a relatively high ISM pressure～107and a high Lorentz factor,it is likely that for a jet to achieve FR2type powers well in excess of1043ergs s?1the ratio of jet to ISM pressures(ξ)is much greater than unity.
3.1.Resolution Constraints on Jet and ISM parameters
3.1.1.Jet parameters
Here we discuss the selection of parameters for the jet,potential and hot and warm components of the atmosphere.Physical objectives dominate the criteria by which we select these parameters but the selection is also governed by the desired resolution and spatial dynamic range constrained by the necessity for a realistic number of cells in the grid.In this subsection we discuss other,global,resolution constraints on the simulation;in the following subsection(§3.2)we consider how the simulations may be scaled.
If a jet is20pc wide where it enters the grid and we require10cells across the jet in order to resolve it adequately with a lagrangian–remap PPM algorithm,then the maximum spatial size of a5123grid is just
over a kiloparsec.Hence,using a code with?xed sized cells,such as ppmlr the range of scales is limited to those relevant to Gigahertz Peak Spectrum(GPS)sources.Nevertheless,some of the features present in the simulations would probably also be relevant to larger scale sources and we indicate these in the sections below.
In adiabatic simulations the precise choice of parameters is not highly constrained since arbitary spatial, velocity and density scales may be applied.However,since we have introduced cooling processes the scaling that is allowed is restricted to a one parameter set(see3.2)so that we need to exercise some care in selecting parameters that provide both physically consistent and interesting simulations,which reveal the range of feasible interactions between a jet and an inhomogeneous interstellar medium.The process of choosing realistic parameters is both interesting and informative.
For the jet the relevant relativistic parameters are:Velocity in units of the speed of lightβ(equivalently Lorentz factorΓ),pressure p jet de?ned by its ratioξto the external ISM pressure and the proper density parameterχ=ρjet c2/4p jet.The non-relativistic counterparts are the velocity v,Mach number M,the pressure,and the ratio of jet to ISM densities,η.
The kinetic power of a jet[see equation(21)]provides one constraint on jet parameters.Another useful constraint comes from consideration of the hot-spot advance speed.Many FR2radio lobes display broad structure that indicates that the head of the lobe is not expanding at high velocity;this is con?rmed by statistical estimates of average lobe expansion in powerful sources～a few percent of the speed of light (Scheuer1995).(See also Blundell et al.(1999).)Following the idea of Scheuer(1982)which more recently has become formally expressed in the form of self-similar evolution of radio galaxy lobes(e.g.Falle(1991); Begelman(1996);Bicknell et al.(1997))the lobe advance is a?ected by the spread of the jet momentum over an area that is larger than the jet cross-section.On the other hand the“instantaneous”hot-spot velocities in a number of GPS sources have been observed to be a considerable fraction of the speed of light(e.g.Conway (2002)).Therefore in the?rst instance it is useful to consider the hot-spot advance speed cβhs calculated from ram pressure balance at the jet terminus.For a relativistic jet expanding into a thermal medium,the instantaneous hot spot advance speed is given by:
βhs β≈
αΓ
1+αΓ
,(23)
where
α=
ρjet+4p jet/c2
ρism
1/2
=
4p jet
ρism c
1/2
(1+χ)1/2.(24)
(Safouris et al.2006).
Let n ism be the total number density of the interstellar medium in the vicinity of the hot spot and let D be the jet diameter.Then,the jet kinetic power can be expressed in terms of the hot-spot parameters as
follows:
F E=μmc3
1+Γ
Γ?1
χ
1+χ
β2
hs
/β2
1?β2
hs
/β2
n ismβA,
≈2.1×1046
1+Γ
Γ?1
χ
1+χ
β2
hs
/β2
1?β2
hs
/β2
n ismβ
D
10pc
2
ergs s?1.(25)
An advantage of this expression is that the precise value of the parameterχ=ρjet c2/4p jet is not important for moderate to high Lorentz factors.Most of the dependence of the jet power enters through the local interstellar medium density,the jet diameter and the instantaneous hot spot advance speed.
At a kiloparsec from the core,D～10pc and n ism～1are reasonable?ducial values.At10kpc, D～1kpc and n ism～10?2?10?3may be more appropriate.In the?rst case,βhs≈0.5andβ≈1 give a jet power≈7×1045ergs s?1typical of the most powerful FR2jets;βhs≈0.03implies a power ～2×1043ergs s?1at the lower end of the FR2range.On the other hand,for FR2jets on a scale of10 kpc,βhs≈0.03,D≈1kpc and n ism≈0.01cm?3gives a power～2×1045ergs s?1.Thus FR2jets remain powerful from small to large scales if the e?ective diameter of the jet widens,in response to the evolving dynamics of the lobe(which con?nes the jet)and the instabilities in the jet itself leading to jittering and ?lamentation.
We have selected a jet for which the instantaneousβhs is of order0.05-0.1and the initial jet diameter is20pc.These parameters,together with a central ISM p/k=106cm?3K places the jet at the low end of FR2power≈3×1043ergs s?1.This choice of parameters produces a jet which initially interacts strongly with the ambient ISM but whose morphology at later times is similar to a classical FR2radio source.Test simulations(not presented here)show that with higher-powered jets the hot spot advance rapidly becomes focused when the jet emerges from the inhomogeneous region surrounding the core and the advance speed is quite high.Thus,this simulation is designed to show the characteristics of the range of interactions that can occur.However,for these characteristics to be be manifest in higher powered sources,it is probably necessary to relax the assumption of an equilibrium disk for the inhomogeneous ISM.Future parameter studies will be used to map out di?erent regimes in more detail.
Radio galaxies are mainly associated with the high luminosity end of the elliptical galaxy distribution. However,FR2galaxies are less optically luminous than FR1s(Owen&Ledlow1994).Therefore,we adopt a velocity dispersion of the baryonic matter of200km s?1.The velocity dispersion of the dark matter is 400km s?1.The latter choice is driven by the physical requirement that the dark matter in an elliptical galaxy is more extended than the baryonic matter and by the numerical requirements imposed by resolution and number of cells discussed above.This combination of di?erent dark matter and baryonic matter scales is naturally treated by the double isothermal potential described in§2.1.
3.1.2.Hot ISM parameters
The selection of parameters for the hot medium is straightforward.We choose a temperature for the (isothermal)medium close to the virial temperature de?ned by the dark matter.As noted above we specify the density by a value of p/k=106cm?3K.The hot ISM distribution is de?ned?rst with the calculated density and pressure applied to each cell in the grid.An algorithm for the warm medium is then applied as described in the following paragraph;this algorithm replaces some of the cells in the hot ISM by warm gas.
The warm inhomogeneous interstellar medium is prescribed in the form of a turbulently supported disk.The aim is to realize a single instance out of the ensemble of possible distributions described by the log-normal,power-law energy spectrum distributions with a spatially dependent mean density described in §2.3.
1.The disk is taken to be isothermal,with a temperature of T w=104K,characteristic of the equilibrium
temperature in an ambient radiation?eld.In order that a disk have a scale height that guarantees signi?cant interaction with an emerging jet,the assigned turbulent line of sight velocity dispersionσt is supersonic.Observationally,this is justi?ed by the inference of supersonic turbulence in the disk in the center of M87(Dopita et al.1997)and the gaseous disk in NGC7052(van der Marel&van den Bosch1998).The density scale of the log-normal distribution at the centre of the galaxy,ρw(0,0,0)
is de?ned by approximate total(thermal plus turbulent)pressure equilibrium with the ambient hot
medium,that is,
ˉρw(0,0,0)
kT
ˉm
+σ2t
≈p hot(0,0,0).(26)
2.The density scale of the warm gas at each cell of the simulation is de?ned by equation(15)for the
mean densityˉρw(r,z)of warm turbulent gas in the potential well of the galaxy.Let fρ(i,j,k)represent the log-normal unit mean cube computed as described in§2.3.1,with the triplet(i,j,k)representing the cell indices.The density of warm gas assigned to each cell is:
ρw(i,j,k)=ˉρw(x c,y c,z c)×f(i,j,k),(27)
where(x c,y c,z c)are the zone-centred coordinates corresponding to(i,j,k).
3.There is a cuto?applied to each cell:Where the statistically distributed density falls below the ambient
hot ISM density the hot ISM in that cell is not replaced.
3.2.Scaling
With adiabatic simulations,the choice of the spatial,velocity and density scales is arbitrary.However, the introduction of cooling(in the thermal gas)restricts the allowable scaling to a one-parameter set.
We de?ne scaling parameters and scaled variables,denoted by primes,through the following relation-ships.All variables have their usual meanings withρbeing the gas density,ρ?the stellar plus dark matter density,ρ2Λρ(T)is the volume emissivity due to cooling(withΛρ(T)the density-based cooling function) andφG the gravitational potential.
x i=x0x i,t=t0t ,
ρ=ρ0ρ ,p=p0p ,T=T0T ,
φG=φG,0φ
G ,Λρ(T)=Λ0Λ ρ(T0t ).
(28)
Scaling the continuity and momentum equations in such a way that the form of the equations is pre-served,is straightforward and identical to the adiabatic case,resulting in the scaling relationships and scaled equations:
1.General scaling,
v0=x0
t0
0=p0=ρ0v20φG,0=v20.(29)
2.The ideal equation of state p=ρkT/ˉm implies that
kT0
=v20.(30) 3.The scaled internal energy equation:
d dt ?h
dρ
dt
=?
Λ0ρ20t0
ρ 2Λ ρ(T0T ).(31)
4.The scaling parameter for the cooing function:
Λ0=
p0
ρ20t0
=
v30
ρ0x0
.(32)
5.Finally,the scaled form of the potential equation for the gravitational potential is
? 2φ G=4πG ρ ?,(33)
where G is de?ned in terms of Newton’s constant of gravitation,G,and the gravitational mass density scale,ρ?,0by
G =Gρ?,0x20
v20
=
GM0,?
x0v20
,(34)
where M?,0=ρ?,0x30.
In order that a given simulation describe a set of scaleable physical situations it is necessary that the scaled cooling function have the same functional form.Thus unlessΛ ρ(T0T )has some special from(e.g.
a power-law)it is necessary that the parameter T0be invariant under scaling.Hence,the parameter v0is invariant;this is the major di?erence from adiabatic scaling in which v0may be arbitrary.The spatial scale x0is arbitrary,but t0=x0/v0is restricted by the constancy of v0.
Moreover,in order that the primed equations describe the same situation,the parameterΛ0must be invariant.Referring to equation(32)this means that the gas density scale is inversely proportional to the spatial scale:ρ0∝x?10.
Invariance of the same scaled gravitational equations requires G to be invariant so thatρ?,0∝x?20. This means that the ratio of gas to gravitating density scalesρ0/ρ?,0∝x0.Hence,if we increase the physical scale of a simulation the ratio of physical gas to gravitating mass density increases.Clearly we cannot do this inde?nitely since this would invalidate the neglect of self-gravity of the gas.
To summarize a given simulation de?nes a one-parameter family of simulations where the scaling pa-rameters satisfy the following constraints:
x0=arbitrary,v0=?xed,t0=x0/v0,
ρ0∝x?10,p0=ρ0v20,kT0/ˉm=v20,
ρ?,0∝x?20,M?,0∝x0.
(35)
with the proviso that x0cannot increase to the extent that the gas density exceeds the density of gravitating matter.
Notwithstanding this restricted one-parameter scaling,the allowable set of physical models allowed by this scaling describes an interesting variety of di?erent physical situations.
3.3.Summary of simulation parameters
3.3.1.Model A
Three simulations were performed in this paper,models A,B,and C.However,the?rst(Model A) is presented as the standard model,in the greatest detail.Models B and C are comparison simulations designed to highlight the e?ect of the density and the distribution(smooth or fractal)of the warm gas.
Tables3.3.2,3.3.2and3.3.2summarize the parameters used in Model A,for the jet,the potential and the interstellar medium respectively.The original scaling uses a spatial scale of x0=1kpc,which is the size of the grid;the parameters relevant to this scaling are given in the third column of the table.However,as shown above there is a one-parameter degree of freedom in the scaling that may be used and indicative sets of parameters are given for x0=0.2and5kpc in the fourth and?fth columns respectively.
3.3.2.Models B and C
In addition to the main simulation model A,models B and model C were performed to look for changes in the interaction sequence with a change in two disk ISM parameters,density and uniformity.
The only physical parameter that is di?erent in Model B is the mean density of warm gas.The central value in model B is20cm?3compared to10cm?3in model A.The third simulation,model C,has the same mean density as model B,but has a perfectly smooth,non-fractal distribution of warm gas.The purpose of model B is to examine the e?ect of the mean density of identically distributed gas.The purpose of model C is to examine the e?ect of the porosity of the gas distribution.In models A and B the distribution of warm gas is such that the jet plasma can force its way through low density channels.In model C the only route for the jet to force its way into the hot interstellar medium is for it to push the dense gas out of the way.
The resolution of Models B and C is a factor of two lower,i.e.256×256×256.However,the lower resolution does not appear to a?ect the comparison.We do not expect the much more uniform density distribution in model C to su?er from lower resolution;in model B we observe similar behavior to that of model A and the jet is adequately resolved with10resolution elements across its diameter.
4.Code and Algorithm Physics
Our simulations use a non-relativistic Piecewise Parabolic Method(PPM)code based on code provided by J.Blondin and colleagues via the web-site https://www.doczj.com/doc/1410919553.html,/pub/VH-1/.We have already commented on the non-realtivistic aspects of the code in§3.
The code has been extensively reorganized for e?ciency and parallel execution on the SGI Altix computer operated by the Australian Partnership for Advanced Computation.We have also added subroutines to advect passive scalars,which track the evolution of di?erent gases and to update the energy density,using an implicit method,when optically thin radiative cooling operates.A cooling function has been implemented which is based upon output from the MAPPINGS shock and photoionization code(Sutherland&Dopita 1993;Sutherland et al.2003).The cooling treatment has been extended for the present models by computing an X-ray spectrum for each temperature point in the thermal cooling function,and using the spectra to construct hard and soft X-ray maps of the thermal gas,as the simulation proceeds.This is motivated by the need to improve the correspondence between simulation and observation,and to calculate more directly observable output variables,compared with the less amenable hydrodynamical variables such as density or pressure.
We have also added code to deal with numerical instabilities that can occur in strong shocks,especially when cooling is present(Sutherland et al.2003).This code is simply referred to as ppmlr for Piecewise Parabolic Method Lagrangian Remap.An important advantage of this PPM algorithm is the relatively low di?usions it exhibits,compared to other?nite volume method for example,allowing the number or cells
Scaled to x0=
Parameter&units Symbol1.0kpc0.2kpc5.0kpc
Equivalent Relativistic Jet Parameters
?Lorentz factorΓ555
?Rest energy density/enthalpyχ101010
Velocity/Speed of lightβ0.97980.97980.9798
Hydrodynamic Jet Parameters
Pressure/External pressureξ 1.0 1.0 1.0
Density/External densityη2.0×10?32.0×10?32.0×10?3
Mach number M25.925.925.9
?Diameter(pc)D jet408.0200.0
Kinetic luminosity L jet2.77×10435.54×10421.385×1044
Assigned parameters are indicated with a?symbol;others are derived.
Table1:Standard Jet Parameters of Model A.
Scaled to x0= Parameter and units Symbol1.0kpc0.2kpc5.0kpc
Double isothermal potential
?Dark matter to
Baryonic velocity dispersionκ222
?Dark matter to
Baryonic core radiusλ101010
Central Baryonic density to
Dark matter density 252525
Dark matter values
?Velocity dispersion(km s?1)σd400400400
?Core radius(kpc)r d 3.50.717.5
Central density(g cm?3)ρd,c1.47×10?223.68×10?215.88×10?24
Enclosed Masses,at r=r d
Dark Mass(M )M d(r d)8.14×10101.63×10104.07×1011
Baryonic Mass(M )M b(r d)4.56×10109.11×1092.28×1011
Total Mass(M )M T(r d)1.27×10112.54×10106.35×1011
Baryonic/Dark Mass ratio M b(r d)/M d(r d)0.60.60.6
Baryonic values
Velocity dispersion(km s?1)σb200200200
Core radius(kpc)r b0.350.07 1.75
Central density(g cm?3)ρb,c3.68×10?219.20×10?201.47×10?22
Enclosed Masses,at r=r b
Dark Mass(M )M d(r b)3.08×1086.15×1071.54×109
Baryonic Mass(M )M b(r b)4.72×1099.44×1082.36×1010
Total Mass(M )M T(r b)5.03×1091.01×1092.51×1010
Baryonic/Dark Mass ratio M b(r b)/M d(r b)15.315.315.3
Assigned parameters are indicated with a?symbol;others are derived.
Table2:Standard Potential Parameters of Model A.
Scaled to x0=
Parameter&units Symbol1.0kpc0.2kpc5.0kpc
Hot Atmosphere:
?Virial/Gas temperatureβh 1.0 1.0 1.0
Gas Temperature(?K)T h1.20×1071.20×1071.20×107
Central Values:
?pressure/k(cm?3?K)p h,c/k1.00×1065.00×1062.00×105
pressure(dynes cm?2)p h,c1.38×10?106.90×10?102.76×10?11
number density(cm?3)n h,c8.35×10?24.17×10?11.67×10?2
mass density(g cm?3)ρh,c8.64×10?264.32×10?251.73×10?26
Warm Disk–ISM:
Virial/Gas temperatureβw1200.01200.01200.0
?Gas Temperature(?K)T w1.0×1041.0×1041.0×104
?Turbulent dispersion(km s?1)σt40.040.040.0
?Rotational Support E R0.930.930.93
Internal Non–Uniformity:
?Log-Normal Meanμ 1.0 1.0 1.0
?Log-Normal Varianceσ2 5.0 5.0 5.0
?Density power-lawβ5/35/35/3
Volume of warm gas(pc3)V w2.55×1072.04×1053.19×109
Mass of warm gas(M )M w4.67×1051.87×1041.17×107
Relative Disk Mass 1.00.0425.0
Central Values:
pressure/k(cm?3?K)p w,c/k1.00×1065.00×1062.00×105
?number density(cm?3)n w,010.050.0 2.0
mass density(g cm?3)ρw,01.04×10?235.18×10?232.07×10?24
Assigned parameters are indicated with a?symbol;others are derived.
Table3:Standard Hot Halo and warm disk–ISM Parameters Model A.
needed to capture shock structures to be minimized,important in3D simulations,and the speed of the algorithm allows the calculation of uniformly high resolutions where more complex algorithms,such as a full MHD treatment may not be practical.
The neglect of magnetic?elds is justi?ed in the?rst instance because of the large increase in simulation phase space that this entails.For example,some simulations incorporate an initially toroidal,unidirectional jet magnetic?eld;others inject a random?eld.The negelect of magnetic?eld probably does not have a major e?ect on the evolution of the simulation for the following reasons.Much of the?ow in these simulations is turbulent and a completely disordered magnetic?eld behaves as a polytropic gas with a ratio of speci?c heats,γ=4/3.This is not too di?erent from theγof5/3that we use here so that we expect a turbulent magnetic ?eld to track the gas pressure reasonably well.Of course there are quali?cations to this.Magnetic?elds and ideal gases behave di?erently in shocks and a systematic component of magnetic?eld may introduce di?erent dynamics than an isotropic turbulent?eld.Moreover,in order to investigate important Faraday rotation and polarisation e?ects as well as the details of the nonthermal emissivity,it is essential to include a magnetic?eld.These e?ects are ones that we can investigate with MHD simulations.Nevertheless,for now,the current simulations provide us with a base for future work in which magnetic?elds are included in a systematic way.Moreover,the general features of the?ow–obstruction by clouds,the initial formation of bubbles and the formation of radiative shocks are probably well captured by the simulations that we present.
We have implemented the following boundary conditions in these simulations:For most of the left hand boundary plane(x=0)the boundary is re?ecting.The exception is the jet inlet y2c+z2c 5.Results In this section we present the results of Model A in the form of multi-panel snapshots from signi?cant epochs;these montages are designed to bring out the relevant physics of the simulations.Some snapshots correspond to slices of important dynamical variables such as density and pressure;in some cases we also present projected versions of variables such as the density.We also present volumetric ray-traced projected images of the radio and X-ray emissivity to produce synthetic images of radio and X-ray surface brightness. 5.1.Evolutionary phases In all cases the times chosen for the sequence of snapshots corresponds to the following phases in the simulation.The mid-plane density slices(Figure2)most clearly illustrate the phases enumerated here,with the other representations highlighting some speci?c facets. 1.Flood and Channel phase:Snapshots at5,10,15and25kyr represent the time over which the jet is making its way through the porous,fractal disk.The shape of the interacting region is amorphous, and determined by?ow of hot high pressure gas along weaker line in the dense disk medium.The pressure in this phase is very high,and the X-ray emission is a strong function of time,depending on a combination of the amount of disk material that is advected and the amount of energy that is processed by radiative shocks,which increases with the size of the region.It is not not well determined what the time dependence of thermal luminosity is in this phase,but it is de?nitely super-linear. In the next two phases an energy-driven bubble forms and evolves,although the beginning and end of this evolution exhibit some di?erent characteristics. 2.Energy-driven bubble phase:Snapshots at35,45and55kyr represent the epoch during a high pressure, pseudo-spherical bubble forms and grows larger than the disk.The jet is still disrupted by disk material, some of which has been advected to higher altitudes,but the bulk of the energy?ux drives the expansion of the nearly adiabatic bubble.A corresponding drop in the e?ciency of conversion of jet energy to thermal emission is seen at about20kyr.Most of the bubble grows(outside the disk)with a power-law that is consistent with classical energy bubble theory.The?ood and channel behavior persists somewhat within the plane of the disk,but this involves an ever decreasing fraction of the jet energy ?ux. 3.Jet-breakout phase:In the epochs represented by55,65,and70kyr the jet starts to break fee of the few remaining clouds in its path and the jet terminus starts to propagate towards the edge of the bubble.The bubble has a generally low density,so that once clear of dense material and re-collimated, the jet transits the bubble quickly. 4.Classical phase:At75,85,and95kyr and beyond the jet pierces the original bubble and then starts to form a classical radio lobe with hotspot,cocoon,bow-shock and back?ow.The remnant of the spherical bubble continues to grow,and?ow continues within the disk,with many radiative and non-radiative shocks throughout the whole disk persisting to late times in all locations bar the very centre cleared by the main jet. 1、SDN是什么？ SDN(Software Defined Network)即软件定义网络，是一种网络设计理念。网络硬件可以集中式软件管理，可编程化，控制转发层面分开，则可以认为这个网络是一个SDN网络。SDN 不是一种具体的技术，不是一个具体的协议，而是一个思想，一个框架，只要符合控制和转发分离的思路就可以认为是SDN. 2、传统网络面临的问题？ 1）传统网络部署和管理非常麻烦，网络厂商杂，设备类型多，设备数量多，命令行不一致2）流量全局可视化难 3）分布式架构中，当网络发生震荡时，网络收敛过程中，有可能出现冗余的路径通告信息4）网络流量的剧增，导致底层网络的体积膨胀、压力增大；网络体积越大的话，需要收敛的时间就越长 5）想自定义设备的转发策略，而不是网络设备里面的固定好的转发策略 -------->sdn网络可以解决的问题 3、SDN的框架是什么 SDN框架主要由，应用层，控制层，转发层组成。其中应用层提供应用和服务（网管、安全、流控等服务），控制层提供统一的控制和管理（协议计算、策略下发、链路信息收集），转发层提供硬件设备（交换机、路由器、防火墙等）进行数据转发、 4、控制器 1）控制器概述 在整个SDN实现中，控制器在整个技术框架中最核心的地方控制层，作用是上接应用，下接设备。在SDN的商业战争中，谁掌握了控制器，或者制定了控制器的标准，谁在产业链条中就最有发言权 2）控制器功能 南向功能支撑：通过openflow等南向接口技术，对网络设备进行管控，拓扑发现，表项下 发，策略指定等 北向功能：目前SDN技术中只有南向技术有标准文案和规范，而北向支持没有标准。即便如此，控制器也需要对北向接口功能进行支持，REST API,SOAP,OSGI,这样才能够被上层的应用调用 东西向功能支持：分布式的控制器架构，多控制器之间如何进行选举、协同、主备切换等3）控制器的种类 目前市场上主要的控制器类型是：opendaylight （开发语言Java），Ryu(开发语言python), FloodLihgt(开发语言Java)等等 5、opendaylight（ODL）控制器介绍 ODL拥有一套模块化、可插拔灵活地控制平台作为核心，这个控制平台基于Java开发，理论上可以运行在任何支持Java的平台上，从Helium版本开始其官方文档推荐的最佳运行环境是最新的Linux（Ubuntu 12.04+）及JVM1.7+。 ODL控制器采用OSGi框架，OSGi框架是面向Java的动态模型系统，它实现了一个优雅、完整和动态的组件模型，应用程序（Bundle）无需重新引导可以被远程安装、启动、升级和卸载，通过OSGi捆绑可以灵活地加载代码与功能，实现功能隔离，解决了功能模块可扩展问题，同时方便功能模块的加载与协同工作。自Helium版本开始使用Karaf架构，作为轻量级的OSGi架构，相较于早前版本的OSGi提升了交互体验和效率，当然其特性远不仅仅于此。 ODL控制平台引入了SAL（服务抽象层），SAL北向连接功能模块，以插件的形式为之提供底层设备服务，南向连接多种协议，屏蔽不同协议的差异性，为上层功能模块提供一致性服务，使得上层模块与下层模块之间的调用相互隔离。SAL可自动适配底层不同设备，使开发者专注于业务应用的开发。 此外，ODL从Helium开始也逐渐完成了从AD-SAL（Application Driven Service Abstraction Layer）向MD-SAL（Model Driven Service Abstraction Layer）的演进工作，早前的AD-SAL，ODL控制平台采用了Infinispan技术，In?nispan是一个高扩展性、高可靠性、键值存储的分布式数据网格平台，选用Infinispan来实现数据的存储、查找及监听，用开源网格平台实现controller的集群。MD-SAL架构中采用Akka实现分布式messageing。 6、ODL的总体框架 ODL控制器主要包括开放的北向API，控制器平面，以及南向接口和协议插件。北向API 有OSGI和REST两类，同一地址空间应用使用OSGI类，而不同地址空间的应用则使用REST 类。OSGI是有状态的连接，有注册机制，而rest是无状态链接。上层应用程序利用这些北 本文作为码农学ODL系列的SDN基础入门篇，分为两部分。第一部分，主要讲述SDN是什么，改变了什么，架构是什么样的，第二部分，简要介绍如何去学习SDN。 1.什么是SDN SDN(Software Define Network) ，即为软件定义网络，可以看成网络界的操作系统。从SDN的提出至今，其内涵和外延也不断地发生变化，越来越多的人认为“可以集中控制、开放可编程和转控分离的网络”就是SDN网络，并且还延伸出软件定义计算、软件定义存储以及软件定义安全等。SDN加快了新业务引入的速度，提升了网络自动化运维能力，同时，也降低了运营成本。SDN的基础 知识如下图所示，下面各小节内容将根据该图内容进行展开论述： 1.1.SDN基础 1.1.1.SDN本质及核心 我们知道，传统网络中的路由器也存在控制平面和转发平面，在高端的路由器或交换机还采用物理分离，主控板上的CPU不负责报文转发，专注于系统的控制；而业务板则专注于数据报文转发。所以路由器或交换机内的控制平面与转发平面相对独立又协同工作，如图所示： 但这种分离是封闭在被称为“盒子”的交换机或路由器上，不可编程；另一方面，从IP网络的维度来考虑，采用的是分布式控制的方式：在控制面，每台路由器彼此学习路由信息，建立各自的路由转发表；在数据面，每台路由器收到一个IP 包后，根据自己的路由转发表做IP转发； IP网络的这种工作方式带来了运维成本高、业务上线慢等问题，并越来越难以满足新业务的需求，传统上通过添加新协议、新设备等手段来缓解问题的方式，收益越来越少。穷则思变，许多人产生了革命的想法，现有的网络架构既然无法继续演进发展，为何不推倒重来，重新定义网络呢？真可谓“时势造英雄”，2006年斯坦福大学Nick McKeown教授为首的研究团队提出了OpenFlow的概念用于校园网络的试验创新，后续基于OpenFlow给网络带来可编程的特性，SDN （Software Defined Network）的概念应运而生。 SDN将原来封闭在“盒子”的控制平面抽取出来形成一个网络部件，称之为SDN 控制器，这个控制器完全由软件来实现，控制网络中的所有设备，如同网络的大脑，而原来的“盒子”只需要听从SDN控制器的命令进行转发就可以了。在SDN 的理念下，所有我们常见的路由器、交换机等设备都变成了统一的转发器，而所有的转发器都直接接受SDN控制器的指挥，控制器和转发设备间的接口就是OpenFlow协议。其简单模型如图所示： OpenDaylight与Mininet应用实战之复杂网络验证（五） 1多交换机的测试 Mininet中本身就支持多交换机网络拓扑的模拟创建，可通过Python API自定义拓扑创建满足使用者在仿真过程中的多方位需求。 下面举出具体示例验证多交换机支持： 执行此条命令后，查看ODL的Web界面显示的网络拓扑。界面拓扑显示如下： 对所有的虚拟host之间进行互ping操作，通过pingall命令，验证主机间的连通性，继而可验证支持多交换机的功能。 由pingall显示的结果可看出，主机间能够互相通信，且将数据包的流转发给交换机，并由交换机上报给ODL控制器来下发流表使主机通信。 主机通信过程中可查看交换机的流表信息及本身信息。 由交换机流表信息显示可知，控制器通过策略将流表下发到交换机中，使主机发出的数据包转发到下一目的地址。每个交换机查看信息的端口都不同，从第一个交换机端口号为6634开始，以后每一个交换机依次在之前交换机端口号的基础上加1，如第二个交换机的端口为6635。其他交换机的流表信息及自身设备信息可根据此说明进行查看。 2多控制器的测试 多控制器验证支持测试包括两种情况： ■OpenFlow网络中多个同一类型的控制器； ■OpenFlow网络中多个不同类型的控制器； 2.1多个同一类型的控制器验证 测试OpenFlow网络中多个同一类型的controller，比如OpenDaylight，多个ODL之间通过 OpenFlow1.0协议标准交互。 通过Mininet验证，在Mininet中模拟创建的OvS交换机不能指定连接多个控制器，且在同一个Mininet中创建的多个交换机不能指定不同的控制器。所以在验证交换机被多个同一类型的控制器管控时，不能通过用Mininet来验证，但是可通过真实交换机来验证。 如，在真实交换机中设置连接此文中的ODL控制器及另一个ODL控制器，命令为： 连接两个相同类型的ODL控制器，其中192.168.5.203为上述实验使用的控制器，192.168.5.111为另外安装使用的ODL控制器。通过执行如下命令查看交换机连接的控制器信息。 is_connected:true表示交换机都成功连接上控制器。交换机连接到这两个控制器后，控制器通过设备拓扑管理也可以发现此交换机，同时控制器管控存在主备关系，但控制器都可对交换机进行管控、下发流表等操作。 通过真实OpenFlow交换机连接多个控制器，可以实施，且已经验证，控制器和控制器之间存在主备关系，多控制器都可以对连接的交换机进行管控。 2.2多个不同类型的控制器验证 在OpenFlow网络中多个不同类型的controller，比如同时存在NOX和ODL，它们之间如果遵循OpenFlow协议标准的话，也是能够协作工作的。多个不同类型的控制器管控交换机与2.1小节是同样的道理。 如，在真实交换机中设置连接此文中的ODL控制器及其他另一个不同类型的控制器，如POX，命令为： 连接两个不同控制器，其中192.168.5.203为上述实验使用的控制器，192.168.5.111为另外安装使用的POX控制器。经试验验证，ODL与POX都遵循OF1.0版本的协议标准，所以在复杂网络多控制器情况下，只要控制器遵循相同的标准规范，控制器之间可进行对网络的通信管理等。此处实验结果与2.1节一致。交换机连接这两个控制器后，控制器管控存在主备关系，但控制器都可对交换机进行管控、下发流表等操作。 3总结 本文主要对复杂网络多交换机及多控制器的支持验证。因Mininet现在无法模拟多控制器管控一个交换机的情况，所以本专题还是侧重对多交换机的管控实验。至此，OpenDaylight 与Mininet应用实战专题将结束，有介绍不到位或者有疑问的地方请多多指教，互相交流。谢谢！ 菜鸟水平初步设置OpenDaylight OVSDB + Openstack测试环境 Hannibal （SDNAP首发） 刚接触SDN和OpenDaylight两个多月时间，还处于人云亦云照葫芦画瓢的水平，在很多大牛的指导文章帮助下，初步搭建一个很简单的OpenDaylight OVSDB + Openstack调试环境。第一次写技术文章，请多包涵。 一、准备 硬件： 双核Core i7，内存4GB，一个以太网卡的Thinkpad X201t，普通个人用笔记本 Host环境： 64位Ubuntu 13.10，OVS 2.0.90 VM环境： 2个Virtualbox VM，Fedora 19 + OVS 2.0.0 + Devstack 。导入Virtualbox都是缺省配置。两个VM的下载地址： https://https://www.doczj.com/doc/1410919553.html,:443/v1/96991703573236/imgs/Fedora19--2node-Devstack.tar.bz2 Size: 4983728003 bytes MD5sum: dfd791a989603a88a0fa37950696608c 二、原理 OpenDaylight（ODL）是一个由Linux基金会支持，多个网络厂商参与的开源SDN控制器项目。Openstack是开源的IaaS项目。如何让两个平台整合以便更好的发挥作用是本环境搭建的目的。 现有的解决方案之一，就是利用Openstack Neutron的ML2 Plugin，将网络复杂性丢到ODL。也就是说，Openstack通过ML2 Plugin，与OpenDaylight的NB API进行会话，具体网络部署的实现交由OpenDaylight Controller来实现。 Openstack的Ocata版本与opendaylight 的Carbon版本集成详解 作者：胡章丰，zfhu2001@https://www.doczj.com/doc/1410919553.html, 前提条件 ===================================================================== 1.已搭建好的可用openstack ocata环境一套 2.已下载的opendaylight carbon-sr1发布版本 3.本文档所述环境地址：控制节点：192.168.137.101，网络节点192.168.137.101，计算节点：192.168.137.101,192.168.137.102，ODL控制器节点：192.168.137.100 4.建议ODL控制器节点与Openstack控制节点采用独立节点安装，否则会有端口冲突，需要修改若干配置文件来避免冲突 ===================================================================== 部署opendaylight控制器 ===================================================================== ODL控制器节点执行： 解压缩软件包 tar xzvf distribution-karaf-0.6.1-Carbon.tar.gz cd distribution-karaf-0.6.1-Carbon/ 开启iptables规则（建议将下列规则写入脚本文件，配置开机自动执行，否则每次重启后需要手动添加这些规则） iptables -I INPUT -p tcp --dport 8181 -j ACCEPT iptables -I INPUT -p tcp --dport 8080 -j ACCEPT iptables -I INPUT -p tcp --dport 6640 -j ACCEPT iptables -I INPUT -p tcp --dport 6653 -j ACCEPT 启动odl控制器 ./bin/karaf 安装odl组件（只能装这几个） feature:install odl-netvirt-openstack odl-dlux-core odl-mdsal-apidocs 验证是否安装成功（打开如果是黑板一块，则说明安装成功） 看看能否打开http://ODL控制器节点ip地址:8181/index.html ===================================================================== OpenDaylight的Helium版本安装 OpenDaylight（后面缩写ODL）的Helium（氦）版本已发布，具体详情可参考ODL官网。Helium（氦）版本只发布了一个版本，下载链接地址为https://www.doczj.com/doc/1410919553.html,/software/downloads/helium。官网中分别共享了版本、安装向导、用户向导、开发者向导手册，可进行下载学习。 git clone https://https://www.doczj.com/doc/1410919553.html,/gerrit/p/integration.git 1.1.Helium安装 此Helium(氦)版本安装研究是基于Ubuntu12.04的基础上进行安装的，此ODL源文件版本是完全可移植的，但是需要Java7.0以上兼容JVM来运行。如果是用到Oracle的话，JDK版本在 1.7.0_45以上。 解压已获取的安装包文件，并进入解压目录： #unzip distribution-karaf-0.2.0-Helium.zip #cd distribution-karaf-0.2.0-Helium/ #cd bin #./karaf##出现问题？ 经验证，此时执行./karaf时，会出现线程异常且No route to host错误，，需要进入上级目录修改文件org.apache.karaf.management.cfg： #cd.. #cd etc #vi org.apache.karaf.management.cfg#打开此文件 将 serviceUrl= service:jmx:rmi://0.0.0.0:${rmiServerPort}/jndi/rmi://0.0.0.0:${rmiRegistryPort}/karaf-${karaf.na me}修改成 serviceUrl= service:jmx:rmi://127.0.0.1:${rmiServerPort}/jndi/rmi://127.0.0.1:${rmiRegistryPort}/karaf-${kar https://www.doczj.com/doc/1410919553.html,}， 再次进入ODL启动目录： #cd bin #./karaf##执行karaf文件 出现以下正确界面，如图所示： OpenDaylight初步学习过程 ———————Lithium OpenDaylight搭建环境的要求 1.虚拟机Ubuntu 14.04，内存建议4G及以上，以免在启动ODL时太卡 2.Java7-及以上版本 3.Maven3.1.1及以上版本 注意： 先用java –version查看jdk版本。如果版本低于jdk1.7，则从jdk官网下载，下载地址：https://www.doczj.com/doc/1410919553.html,/technetwork/java/javase/downloads/java-archive-downloads-javase7-521261.html#jdk-7u79-oth-JPR一定要根据自己系统下载相应的jdk。 安装及配置：https://www.doczj.com/doc/1410919553.html,/s/blog_93dc666c0101b1bj.html 查看maven版本，maven –v，如果未安装，则从其官网下载3.1.1版本及以上版本。 Tar文件，可以先去官网查下maven最新版本多少。 下载网址：https://www.doczj.com/doc/1410919553.html,/dyn/closer.cgi/maven/binaries/apache-maven- 3.3.3-bin.tar.gz 安装配置：https://www.doczj.com/doc/1410919553.html,/caojianhua/archive/2011/04/02/347559.html 建议不要从shell通过apt-get来安装maven，版本不是最新的。 安装pre-build的controller 由于新手初期对于ODL的了解还不多，建议先安装pre-built的distribution熟悉一下opendaylight的基本功能。 1)下载地址如下，下载zip格式 https://https://www.doczj.com/doc/1410919553.html,/downloads 2)解压文件，进入到bin文件夹，运行./karaf Md-sal中 How To Look Up Data In MD-SAL –Helium Version Posted by Kanika Previously I wrote how to look up data in MD-SAL data store but that holds good only for OpenDaylight’s Hydrogen release. In OpenDaylight’s Helium release, data broker API’s have been changed. Here is how you can look up data in MD-SAL data store if you are using OpenDaylight’s Helium v ersion. Note that you also have to switch your OSGI bundle to be config subsytem aware before you can start using new DataBroker service. import https://www.doczj.com/doc/1410919553.html,mon.base.Optional; import org.opendaylight.controller.md.sal.binding.api.DataBroker; import https://www.doczj.com/doc/1410919553.html,mon.api.data.LogicalDatastoreType; import org.opendaylight.controller.md.sal.binding.api.ReadOnlyTransaction ; ... ... //Look up all the nodes from MD-SAL operational data store InstanceIdentifier OpenDaylight开发学习笔记基础篇 一、摘要 本文主要针对Openflowjava部分进行实例简述，初学者需要对java了解一些，总结一些我自己的学习收获，不足之处请指正。 Openflowjava工程作为Opendaylight南向接口的协议栈存在，与openflowplugin工程及外部的netty.io网络库紧密联系。其主要作用是接受南向接口上报的消息、解码、将其交给Openflowplugin以便进一步上报以及接收Openflowplugin传达的发送消息的指令并将其编码为字节流从南向接口中发出。结构上与功能相关的是Openflow-protocol-api及Openflow-protocol-impl两个文件夹下的代码。前者中用一系列yang文件定义了控制器支持的Openflow消息结构及对应的收、发行为，后者包含了消息的解码、编码及上报、下发的功能逻辑。 netty.io网络库在Opendaylight中负责处理控制器与switch间的TCP连接。控制器接收上报的Openflow消息时，消息的二进制字节流会被装入netty的Bytebuf类对象中传递给Openflowjava来提供解码，发送消息时，消息相关的数据在编码序列化后被封装进Bytebuf类中交给netty完成发送。Openflowplugin 作为opendaylight处理openflow消息的外挂组件进一步负责解码后消息的分类、处理、上传以及发送消息时控制器指令数据的分发、消息体数据的组织与打包。Openflowplugin控制着openflow消息收发的流程和逻辑，openflowjava作为其末端动作的执行者存在。 二、Openflow-protocol-api部分开发 openflow-protocol-api文件夹下的代码主要是openflow-protocol-api/src/main/yang路径下的一系列yang文件，yang是一种表示结构与属性的语言，同时也是一个RFC标准（RFC6020）。Opendaylight使用这种语言来定义其所支持的openflow消息结构以及消息的收发动作。在项目编译阶段，maven会调用opendaylight的yangtools工具根据yang文件生成一系列java文件，包括各openflow消息的容器类、容器类的构造器以及侦听并接收这些容器类的接口。通过yang文件，我们可以简单清晰地定义我们需要的消息类型以及其上传/下发管道。 OpenDaylight与Mininet应用实战之OpenFlow1.0协议分析继本专题基本环境搭建（一）之后，本文在此基础上熟悉平台操作，以及通过wireshark 抓包工具分析OpenFlow（以下简写为OF）协议。具体的OF官方协议及白皮书可在资料库栏目中下载阅读。注：此文涉及的环境仅支持OF1.0版本，对于OF1.2、OF1.3版本可用其他平台测试，后续会另做专题讨论。 1打开wireshark并创建拓扑 按照章节一搭建平台，启动ODL，并打开wireshark。进入装有Mininet的VM，通过mn命令指定网络拓扑及指定此ODL控制器。 Mininet创建网络拓扑命令： 此命令通过Mininet模拟创建一个含有两个交换机（Open vSwitch，以下简写为OVS）和两个主机的网络拓扑，其中192.168.5.203为ODL的IP，6633为ODL的默认端口，网络拓扑如下图所示： 2查看网络 在Mininet中通过操作网络命令，可以查看OVS间及OVS与主机间的连接关系，也可以查看Mininet是否远程连接控制器。 例如，通过nodes命令可以查看网络中所有的节点。 通过net命令可以查看并确认网络连接关系是否与预期一致以及节点信息，且可以了解具体的连接端口信息。 通过下面的dump命令可以看出，交换机通过远程方式连接到控制器，且能看到控制器的IP 和PORT。 3抓包并分析协议 通过wireshark抓包可以直接看到控制器与OVS交换机的通信过程，下面分析该流程中的OF消息。此专题应用的是直接支持OpenFlow协议的wireshark官网Stable Release(1.12.1)版本。 3.1建立连接 控制器与交换机之间的OpenFlow协议是应用于TCP传输层上，所以解析应用层。他们首先发送hello消息，建立初始化连接，协商使用的OpenFlow协议版本。由下图可知，ODL与Mininet之间应用的是OpenFlow1.0版本协议（其他1.2、1.3协议会在协议OpenFlow 后面标识）。 3.2能力请求响应 该消息主要响应能力请求feature request消息，回复连接此控制器的交换机的一些基本设置信息，包括交换机的能力以及它的一些端口的信息等。 DOM DataStore 修改记录 新标准化树形数据模型 对于数据的新标准化模型将代表后面的YANG 规格的实际概念。这将不再是基于序列化格式（定义见YANG 规格和使用的SAL-经纪参数impl1.0）。 ●NormalizedNode 代表在树结构中的一个节点的基本类型;所有其他节点类型是来自这个基本类型。它包含一个叶子识别符和一个值。 ?DataContainerNode 其中包含多个叶节点;它在YANG语法中没有直接表示。 ◆ContainerNode 容器节点，它包含了多个子叶片和映射到YANG容器陈述。（Node, which represents a leaf which can occur only once per parent node; it contains multiple child leaves and maps to the container statement in YANG.） ◆MapEntryNode 节点代表一个叶子，叶子是唯一识别的键的值。MapEntryNode中可以包含多 个。（Node which represents a leaf, which can occur multiple times; a leave is uniquely identified by the value of its key. A MapEntryNode may contain multiple） ◆ChoiceNode 节点代表一个叶子，这大多发生一次，每个父节点，但可能的值可以有不同的 类型。地图来选择语句。类型映射到该选择的case语句。 ◆AugmentationNode Node which represents a leaf, which occurs mostly once per parent node. ?LeafNode 叶子节点，存放相关值（PS：官网解释是 Contains simple value.）。 ?LeafSetEntryNode ?LeafSetNode ?MapNode Data Store接口 Opendaylight学习文档qq群#北邮-天依 目录 1.概述 (3) 1.1 Opendaylight简介 (3) 1.2本文档组织结构 (7) 2.感受Opendaylight (7) 2.1 环境搭建 (7) 2.2获取代码 (9) 2.3安装mininet (11) 2.4 controller使用及功能介绍 (11) 2.5Openflowplugin功能及使用方法 (14) 2.6 Hydrogen (16) 3 Maven和OSGI基础 (16) 3.1 Maven (16) 3.2 OSGI (20) 4 使用IDE (30) 4.1 使用Eclipse (31) 4.1.1 导入controller项目 (31) 4.2 使用Intellij idea (38) 5Controller代码分析 (39) 5.1 代码目录 (40) 5.2 收发包过程简介（packet service） (41) 6 Opendaylight重要技术及文档 (44) 1.概述 1.1 Opendaylight简介 Opendaylight（Opendaylight官网）是Linux基金会的一个合作项目。目前，包括十二个项目，每一个项目都有自己的代码库(Opendaylight项目列表)。这些项目中与openflow相关的项目的有controller、openflowjava和openflowplugin，目前，controller仅支持openflow 1.0，openflowplugin是一个单独的项目，将来它的core部分要集成到controller中，使controller支持openflow 1.3及以上的版本。Opendaylight的厂商成员分为铂金成员，金牌成员和银牌成员。 图1 Opendaylight阵营 Opendaylight controller使用java编写，运行在JVM上，理论上来说可以部署到任何支持JA V A的平台上，但是其官网文档推荐的最佳运行环境为最新的Linux（Ubuntu 12.04+）及JVM 1.7+。OpenDaylight Controller提供了一个模块化的开放SDN控制器，它提供了开放的北向API（开放给应用的接口），同时南向支持多种包括openflow在内的多种SDN协议。底层支持混合模式的交换机和经典的Openflow交换机。 Open Daylight Controller在设计的时候遵循了六个基本的架构原则（以下来 OpenDaylight REST APIs研究 1 Topology REST APIs 在org.opendaylight.controller.topology.northbound包中。 Topology Northbound REST API Authentication scheme : HTTP Basic，Authentication realm : opendaylight，Transport : HTTP and HTTPS。默认使用HTTPS Authentication 1.1 Topology REST APIs可提供的服务 客户端请求服务器响应 GET：检测拓扑topology GET：检测用户配置连接list PUT：添加一个用户连接topologyUserLinkConfig 1.2 Topology REST APIs数据模型 其REST数据元素及相应的XML配置细节如表4.2-1所示。 Data Elements Data Types XML Elements 名称（类型） 最大/最小出 现 edge edge tailNodeConnector（nodeConnector）0/1 headNodeConnector（nodeConnector）0/1 edgeProperties edgeProperties edge (edge) 0/1 properties/property 0/unbounded list topologyUserLinks userLinks (topologyUserLinkConfig) 0/unbounded node node type (string) 0/1 id (string) 0/1 nodeConnector nodeConnector type (string) 0/1 id (string) 0/1 node (node) 0/1 property property topology topology edgeProperties (edgeProperties) 0/unbounded topologyUserLinkConfig topologyUserLinkConfig status (string) 0/1 srcNodeConnector (string) 0/1 dstNodeConnector (string) 0/1 name (string) 0/1 表4.2-1 Topology REST APIs 基于OpenFlow的SDN技术 1.SDN与OpenFlow技术简介 SDN指的是软件定义网络。你可以这样去理解SDN：如果将网络中所有的网络设 备视为被管理的资源，那么参考操作系统的原理，可以抽象出一个网络操作系统（NetworkOS）的概念—这个网络操作系统一方面抽象了底层网络设备的具体细节， 同时还为上层应用提供了统一的管理视图和编程接口。这样，基于网络操作系统这个 平台，用户可以开发各种应用程序，通过软件来定义逻辑上的网络拓扑，以满足对网 络资源的不同需求，而无需关心底层网络的物理拓扑结构。 云计算的发展，是以虚拟化技术为基础的。云计算服务商以按需分配为原则，为 客户提供具有高可用性、高扩展性的计算、存储和网络等IT资源。虚拟化技术将各种物理资源抽象为逻辑上的资源，隐藏了各种物理上的限制，为在更细粒度上对其进行 管理和应用提供了可能性。近些年，计算的虚拟化技术（主要指x86平台的虚拟化） 取得了长足的发展；相比较而言，尽管存储和网络的虚拟化也得到了诸多发展，但是 还有很多问题亟需解决，在云计算环境中尤其如此。OpenFlow和SDN尽管不是专门 为网络虚拟化而生，但是它们带来的标准化和灵活性却给网络虚拟化的发展带来无限 可能。 SDN的最重要特征：将传统网络设备的数据转发（dataplane）和路由控制（controlplane）两个功能模块相分离，通过集中式的控制器（Controller）以标准化的接口对各种网络设备进行管理和配置，那么这将为网络资源的设计、管理和使用提供 更多的可能性，从而更容易推动网络的革新与发展。 OpenFlow是实现SDN最常用的的一种协议，OpenFlow的原理和基本架构如下图。其实，这张图还很好地表明了OpenFlowSwitch规范所定义的范围—从图上可以看出，OpenFlowSwitch规范主要定义了Switch的功能模块以及其与Controller之间的通信信 道等方面。 OpenFlow几大应用场景，包括：1）校园网络中对实验性通讯协议的支持（如其标题 所示）；2)网络管理和访问控制；3）网络隔离和VLAN；4）基于WiFi的移动网络；5）非IP网络；6）基于网络包的处理。 2.OpenFlow协议 Openflow规范主要分为如下四大部分 1.?OpenFlow的端口（Port） OpenFlow规范将Switch上的端口分为3种类别： a)?物理端口，即设备上物理可见的端口； b)?逻辑端口，在物理端口基础上由Switch设备抽象出来的逻辑端口，如为tunnel或 者聚合等功能而实现的逻辑端口； c)?OpenFlow定义的端口。OpenFlow目前总共定义了ALL、CONTROLLER、TABLE、 IN_PORT、ANY、LOCAL、NORMAL和FLOOD等8种端口，其中后3种为非必需的端口，只在混合型的OpenFlowSwitch（OpenFlow-hybridSwitch，即同时支持传统网络协议栈 和OpenFlow协议的Switch设备，相对于OpenFlow-onlySwitch而言）中存在。 2.?OpenFlow的FlowTable（流表） OpenFlow通过用户定义的或者预设的规则来匹配和处理网络包。一条OpenFlow的规 则由匹配域（MatchFields）、优先级（Priority）、处理指令（Instructions）和统计数 据（如Counters）等字段组成，如下图所示。 在一条规则中，可以根据网络包在L2、L3或者L4等网络报文头的任意字段进行匹配，比如以太网帧的源MAC地址，IP包的协议类型和IP地址，或者TCP/UDP的端口号等。 ODL学习 修改记录 1.1ODL概述 Opendaylight是一个以模块化、可插拔、灵活的、基于Java的控制器为核心的开源平台。从北往南，它首先包括供上层应用和业务逻辑使用的北向开放API，有OSGi和REST两类。上层应用程序利用这些北向API获得网络智能信息、运行算法处理分析以及组合新的网络策略；其次，它包括控制器平台本身，它是一组可动态组合的模块用于汇集网络信息，比如网络中有哪些元素、其统计信息 如何等；最南边是能够支持多种协议的南向接口，如Openflow 1.0 1.3 BGP-LS等，这些南向接口可以调用设备上服务抽象层SAL。 1.2ODL架构原则 Open Daylight Controller在设计的时候遵循了六个基本的架构原则： (1) 运行时模块化和扩展化（Runtime Modularity and Extensibility）：支持在控制器运行时进行安装、删除和服务的更新。 (2) 多协议的南向支持（Multiprotocol Southbound）：南向支持多种协议。 (3) 服务抽象层（Service Abstraction Layer）：南向多种协议对上提供统一的北向服务接口。 (4) 开放的可扩展北向API（Open Extensible Northbound API）：提供可扩展的应API，通过REST 或者函数调用方式。两者提供的功能要一致。 (5) 支持多租户、切片（Support for Multitenancy/Slicing）：允许网络在逻辑上（或物理上）划分成不同的切片或租户。控制器的部分功能和模块可以管理指定切片。控制器根据所管理的分片来呈现不同的控制观测面。 (6) 一致性聚合（Consistent Clustering）：提供细粒度复制的聚合和确保网络一致性的横向扩展（scale-out）。 1.3框架概述 上图所示，南向通过plugin的方式来支持多种协议，包括OpenFlow1.0、1.3，BGP-LS 等。这些模块被动态挂载到服务抽象层（SAL），SAL 为上层提供服务，将来自上层的调用封装为适合底层网络设备的协议格式。控制器需要获取底层设备功能、可达性等方面的信息，这些信息被存放在拓扑管理器（Topology Manager）中。其他的组件，包括ARP handler、HostTracker、Device Manager 和Switch Manager，则为Topology Manager生成拓扑数据。 控制器为应用（App）提供开放的北向API。支持OSGi 框架和双向的REST 接口。OSGi框架提供给与控制器运行在同一地址空间的应用，而REST API 则提供给运行在不同地址空间的应用。 所有的逻辑和算法都运行在应用中。 控制自带了GUI，这个GUI 使用了跟应用同样的北向API，这些北向API 也可以被其他的应用调用。 1.4功能概述SDN及ODL概括性总结
ODL之SDN入门篇
OpenDaylight与Mininet应用实战之复杂网络验证(五)
菜鸟水平初步设置OpenDaylight-OVSDB-+-Openstack测试环境
Openstack的Ocata版本与opendaylight 的Carbon版本集成详解
OpenDaylight的Helium(氦)版本安装
OpenDaylight初步学习过程
opendaylight Md-sal
OpenDaylight开发学习笔记基础篇
OpenDaylight与Mininet应用实战之OpenFlow1.0协议分析
Opendaylight DOM DataStore研究
Opendaylight学习及开发初级教程
OpenDaylight REST API研究
SDN和Openflow详解+mininet与opendaylight环境搭建与测试
OpenDayLight 代码学习研究
相关主题
文本预览