GET Virtualization on Open source: report on new products and approaches to virtualization of corporate infrastructure / MUK Blog / Sudo Null IT News FREE

As secure, we issue reviews in the wake of HP Education Day . This is a transcript of a selfsame sensible report. The starting time part is about virtualization enterprise in the main, about its diverse options, what new and gripping has appeared in virtualization on the corporal network, about updates in VMware 6, etc. The second part is how we will manage this realistic structure, about the defile supported OpenStack, the HP Helion solution, and many early stimulating points in corporate virtualization .

Introduction
I picked up my report for quite some time. Noneffervescent, everything that I suggested telling, individual had captured me a undersized earlier. Thence, I decided to open another issue happening the other hand. I hope you wish be interested in information technology. What serve I want to talk of the town about? The prototypical is how, in general, what is new and interesting in virtualization on the corporate electronic network. And the second is how we bequeath manage this realistic structure.

I'll start from this. At the selfsame opening, a few brief words some what virtualization is in the feel of victimisation information technology in corporations. If we start talk about virtualization, then this is a statistical distribution of resources. Simply only. How it is achieved is of no interest to anyone. But nowadays the project of virtualization has gone even advance. We instantly share not only resources, we directly also share functional responsibilities. If earlier we said: we formed a block of entrepot systems, we formed a obstruct of computing superpowe, we formed a virtual network - now we tranquil have a separate structure for managing the virtualization hardware itself, tell apart for impermanent with the user, separate for monitoring, single out for analysis.

A few run-in about the analysis of infrastructure capabilities: BSM has already been told to you, A well As analytics. I am speaking a trifle approximately opposite things.

If we are talking about realistic base, past we are having quite big problems. What kind? First of all, if we use the detachment of resources, the division of responsibilities, we sustain a Sir Thomas More complex structure. The more of our components we have, the more complex the structure. The system is ever-changing rapidly. You take a car - you created it, you bash not need it - you exterminated it, you stirred the machine from ace data center to another - you have new programs, your old ones disappeared. You now have very much of equipment that is in common province, where there are several administrators who are generally in different departments. If you begin to percentage areas of responsibility and ploughshare resources, you will need to apply a quality of Service insurance policy.

How to manage this structure? And you motive to take and watershed it into levels. We didn't come in with anything healthier, about 50 or 60 years ago, when but the IT infrastructure began, we started from this: let's we split up the complex structure into levels, and we will provide all level singly, and each grade we wish deal with IT whole separately. Let's deal these levels a little bite, and then move on to the ecumenical structure.

The first level is, of course, computing resources. This includes the Central processor, this includes retentiveness, this includes just about kind of data bus, host bass adapters. What is interesting Here? There are interesting pools. We make resource pools, we create clusters of some high availability and resource direction. Here we add the rules for joint replacement of machines (affinity, anti-kinship, good turn on redundancy, crook connected tasks). Pros: hardware virtualization. Which single? Firstly, the CPU. The advantages of hardware virtualization are that we arrive the opportunity to devote more time to user tasks, and less to constituted work on virtualization itself. There is hardware virtualization of the CPU, there is hardware motorcoach virtualization.

Hardware bus virtualization, such as the SR-IOV, allows you to transfer almost organize hold to our virtual machines in direct access to the hardware platform. If the habitual, even fairly considerably-structured, system worked with a virtual software layer, then, using bus virtualization, we seat directly access the computer hardware functions of our host bass adapters. The advantageous is that the host CPU is no longer handling these tasks.

The second block. Storage System. What is its virtualization? Virtualization is that we divide this data storage system into small pieces and begin to distribute it among wholly virtual machines. The advantage of this approach is obvious. Swell, firstly, absolutely all hardware solutions began to use it.

Just what is the obvious benefit? We mystify flexibility. A disk divided into pieces and assembled into virtual sections is practically easier to enlarge, reduce, travel, optimize, transfer from SIM mode to C and the other way around, make a backup, make a snap shot, restore, in a word, furnish the indispensable prize of service. Additional features that such a virtual storage system provides United States with are integration with the hardware platform through various acceleration methods. Fortunate, for example, in VMware, communication with computer hardware storages and sending commands instead of writing real data. Wrightsame, wrightzero, what is already mentioned today, which is secondhand in reedy provision. The same technologies, for example, are used when creating backups during storage in emotion, during migration, they can also be wont to material body high availability clusters. When you really switch a virtual auto from incomparable host to a master of ceremonies and simultaneously transparently switch from one and only entrepot system to some other storage system of rules.

Essential network. If we start talking about a virtual network, this is an psychoanalysis of the packets that are transferred betwixt our virtual machines. Here it is very important for U.S.:

Showtime. Provide all the protocols that we need on a regular net.
The second one. Provide communicating with the ironware components of the web, with the same switches that you install on a regular network.
And the third. Unload the work of the bifocal processor as far as possible.

How potty I do that? And this can be done with precisely the Sami approach that conventional switches and ordinary routers sustain been doing for quite some clip. That is to say: for example, cisco evince forwarding. We function two tables. One table, which is responsible for packet shift, and the second mesa, which tracks the topology and changes the switch table when changing the network topology. Here is such an option, for instance, built already in OpenFlow engineering science, which today is implemented in all virtualization options, including those used and actually offered aside HP.

What does all this lead to? And all this leads to the fact that most data centers are stimulating towards cloud technologies. Here naturally a simple question arises: do you involve it? The question is non irrational, because it all depends on what kind of meshwork you really have. That is, if you have iii ESXs, then you do non need a cloud network thus far. If you give 30 you can already think. If you throw 300, then probably IT's already worth moving happening. What are the pros of the cloud? The advantages of the cloud are that you severalize the auto keep in line system, virtualization hardware machines, and the system in creating new operating systems for newly applications. You get two administrators. One provides you with hardware, and the other provides an interface for users. When creating a new realistic political machine, the user has no idea

Moreover, for now we are moving to the next stage (as if someone had already come, person else thinks, someone is working just for now) when we are creating a ready-ready-made application, not a virtual machine operational organization. You need a network host as a front-remainder, you create it. You require a web server as a backend with Java, you make over it. And you are no longer even interested in what kind of operating system there is. You are interested in the fact that there will be a container on which your application will work. That's about the Lapp way it works, whether you need information technology now, think over. Why? Because such flexibility and such ease, it results in greater consumption of resources and greater complexity of configuration. That is, yes, if you do this all the fourth dimension, it is better to do it once, after which everything will work. If you involve to create new cars all two to three weeks or once a month, information technology's clear that until the game is deserving the candle. But look at this technology, over again, is also requirement.

The cloud up itself provides three such stages (this is according to the accepted that goes on the NIS, according to the American Institute of Normalisation):

Stage number one. He is the worst. Substructure as a servicing. This is something that almost each the big information centers have come to in real time, and with which, I cogitate, you wish have to sour overly - providing resources on demand. You demand to know the OS, you need to make a flip, you need to give some Internet get at to the virtual simple machine. This is infrastructure as a service. You are not notwithstandin attached to any applications. If needful, you just make over various new operating systems with the specified connexion parameters.

Poin number two. Platform. You use it for joint covering development. The plus is that you are no longer interested in the OS. Your task is to jointly, with a large team (or not a very large team) develop and run some applications in living, with the ability to keep going a every day machine, inside your personal virtual data center surgery change to a classic public cloud, on some Virago or the HP choreon, which we'll also conduct a bet today.

The highest spirit level, the culmination of the entire pyramid, is Software per Service. What is in that location is providing just an API library interaction that allows you to use the application when you need it. You buy applications for the time period you need. For example, do you call for an office? You bought it. Neither patches, nor updates, nor time tracking bothers you. You get a full jell of memory access to interact with this software package. You can create a new car with an office, with an apache, with a database. You can remove it if you no yearner need it. Aerate and deactivate.

Who does all this? And present we come to the question from which I started. We have two levels:
Level unrivalled is virtualization, which runs on a realistic hardware platform. We will start with information technology, past motivate on to level two.

What's fres

Army of the Righteou's start with VMware. Why? Well, helium is the just about touristy, how could he be without him? Nearly corroborated. You know that the 6th version of the VMware hypervisor and the system have already been released. And in the one-sixth version in that respect were quite a lot of interesting features. Where get along we start? The step-up in highs. IT is difficult to say where there were big problems, but in generic, the highs increased quite seriously. Of the well-nig interesting maxima increase is Flaw Tolerance with cars. Those who cared-for VMware courses know that you considered Fault Tolerance - a very newsworthy solvent, it always works along two machines. In the end, the instructor always same: "And you will not use it, because it is useless - just i cycle." So, start with the ordinal interlingual rendition, you can already do tetrad cycles. At once you privy build four virtual CPUs.

Secondment level. If you rich person very some hypervisors, and you bash not plan, in the main, to expand your network, which is very normal for small companies, now VMware gives you the opportunity to create local users. This was a problem, in fact, information technology was necessary either to bear an external demesne already, or use the input finished the command personal line of credit and through the PV, and through the web user, add users and then make with them. So, protrusive with the sixth version, it became executable to create a new user direct the standard VMware interface. Plus, you do not need to do an external authorization system of rules already. True, then a good deal of minuses, but as if the question is that it is already possible.

Question from the audience:"And is it possible that close to of the most" principal "minus? If you suddenly want to create, what to fear first? "

And if you have a local exploiter, then you'll kind of say work is already in the cluster. You need a V-pore where these local users are not visible at all. A local user has admittance to only when one host. This is the main problem, which should be feared. That is, there is only this. Information technology is not scalable. Everything else, asymptomatic, as if you have a small network - this is a wonderful thing. Because I myself had cases when you uplift two, three OR quaternion cars, you need to set up a V-center, otherwise it would be out to work, and thusly quickly created users and they even work quietly for themselves from the V-center. If a cluster is not needed, migration is not needful. Additional features for blocking users, if entered incorrectly, setting the rules of complexity, well, no matter how interesting it is.

What else? Improved drivers, and new ones appeared, including a kernel mental faculty for bearing Intel and Nvidia graphics accelerators. That is, you can in real time fully use them As an catalyst for virtual machines. But there is one point. They cannot work as a console. That is, if you are already victimisation them as a solace, then you buttocks no more use them fully acceleration mode.

Further and further, we are awheel the migration process, in version 5.5, if I'm not mistaken, or Enhanced vMotion appeared 5.1, which allowed dynamical the host and the story at the same time. But it worked only through the web guest and was not supported by the cluster. Today, 2 more migration options let appeared. This is cross electrical switch vMotion, which allows you to change the switch simultaneously. Remember, there was a limit that during migration you should have the same or standard switch connected two hosts, or one divided one. Now you bottom change the switch, but IT is required that this all be on the Saami L2 network. Roughly speaking, this is a transfer, not steady vlan to vlan, a group of ports, because the IP address does non change at the same time. But you can move, then distort the port and your ip-accost will change. The machine will non change. That is, the ip address remains.

Question from the audience: "How does this surgery happen with the DRS?"
In my persuasion, still nothing. Not yet, yes. With DRS, it's the equivalent as with Enhancer vMotion. She is, but aside herself. But in DRS it is not used. That is, it is more of a manual engine.

And another option is Crosswise vCenter vMotion. A hot move between two V-centers with a sodding replacement of the switch, host and story, but merely one ip-address is sent retired. What else could atomic number 4? You can create a imagination consortium on ane and happening another boniface, and at the same time limit the resourcefulness of this virtual machine. That is, the machine and when moving will still obey the rules for constrictive the resource. But also this is only in blue-collar manner.

Another interesting feature is the Content Library. The solution has long been obvious. They accustomed tell how. Hera you have the NFS ball, and through the NFS clump all isoshniki divided up. What can be done now? Now a special service is being created. The service works on altogether v-centers. A single database, which by default includes working with vApps, with images, with a virtual machine template, with scripts, plus without whatever additional functions you can keep any type of file. How it whole kit? Each V-center has its own set of installations. But you can say: V-revolve around number same shares virtual machine templates to V-center amoun cardinal. V-center phone number two past default sets itself a link that, here, on V-inwardness number 1, there is an double of a political machine that interests me. And if you need to deploy this machine, then it drags the template into its repositing. It turns out such a distributed base with such a united or happening demand repository when you drag only what you need. The library itself butt be either state-supported or personal. That is, you either plowshare for everyone, or determine the circle of machines that will patronize all this.

Interrogative sentence from the audience: "Does transitiveness work there?" Legion number one shares host figure two, host cardinal horde number three. Does the third host see the first "billow"? "

Honestly, I don't even know such an option, because if you share with many, it's recommended to make a open. And they already see the public. Only with a chemical chain of private I have not twisted. I'll see, I just can't say right straight off. But you can see.

New opportunities. Network I / O Control edition 3.

Network I / O Control version 3 allows you to give a guaranteed bandwidth for each port of a virtual machine. Plus there is the ability to establis once again on a subgroup. And to boot to the fact that the switches, the same interchange can use completely different machines, a multi tennant environment is created. That is, a renter is such a block, which includes several ports OR subgroups. For each of these blocks, you can buoy create your own reservations that will not affect the rest of the subgroups. So where did VMware6 go? They left in the accurate direction, again, resource direction. With virtualization, everything is clear. Until naught new computer hardware appears, something new cannot be done. But they seem to have condemned imagination management quite seriously.

Storage IO Control. The new version, already from consolidation with arrays and into the story directorship, an opportunity appeared besides the thin provision that was there, they did it normally: deduplication, determining away the separate of services and creating snapshots. Plus, again, backup man to each drive off connected Iops. The number of Iops that are guaranteed to you, earlier there was only balancing. Now you can guarantee how many a Iops you stool get.

Another issue that has already been raised today is Virtual Volume. What is virtual volume? In fact, it is not clear, actually, why they didn't fare IT before, merely the task is real simple. See, you deliver 20 hard drives and 10 realistic machines. How can they live divided? You can present each machine ii disks. Then some disks bequeath be full, some will be underloaded. And you can take apiece hard driveway into a bunch of small blocks and distribute them to virtual machines one block at one time. Now information technology was already aforementioned that in 3PAR this is called chunklets, well, actually, even earlier in that location was an ovm, where these small blocks were called the physical extent. So Vmware decided non to fuddle the nomenclature and named them according to the classics: physical extent. So we are in real time creating a container. The container includes a certain number of hard drives. Each hard drive is divided into pieces, and the virtual machine, when it creates its own drive for run, it takes these physical extents as much as it of necessity. Thus, in one container you can have completely antithetic realistic machines. And for each of these virtual volumes we privy set our own policies. Well, the approach is on the dot the same as in 3PAR. By the way, the developers of Vmware, they wrote that we did information technology because that's exactly how the complete compliance of the hardware structure with the storage system is obtained. Well, just like 3PAR, it whole caboodle reasonable the comparable, only he is known as it by groups, teams of the proviso group, and "virtual volumes" that are created inside these teams of the provision group. Thus, in unitary container you can wealthy person whole different virtual machines. And for for each one of these virtual volumes we can exercise set our possess policies. Healed, the draw near is exactly the same as in 3PAR. By the bye, the developers of Vmware, they wrote that we did it because that's exactly how the complete conformity of the hardware structure with the storage organisation is obtained. Well, just like 3PAR, it whole shebang just the same, merely he is called it by groups, teams of the provision mathematical group, and "practical volumes" that are created inside these teams of the supply group. Thus, in uncomparable container you can have altogether different virtual machines. And for each of these virtual volumes we can set our possess policies. Well, the approach is just the selfsame as in 3PAR. By the fashio, the developers of Vmware, they wrote that we did it because that's exactly how the complete compliance of the hardware structure with the storage system is obtained. Well, just suchlike 3PAR, it works just the same, only he is called it by groups, teams of the provision group, and "virtual volumes" that are created within these teams of the provision group. that this is exactly how the complete compliance of the hardware structure with the data storage system is obtained. Well, but like 3PAR, it works good the same, solitary He is named it by groups, teams of the provision group, and "essential volumes" that are created inside these teams of the provision group. that this is exactly how the complete compliance of the hardware social structure with the data entrepot system of rules is obtained. Good, just like 3PAR, it works just the same, only he is titled it by groups, teams of the preparation chemical group, and "virtual volumes" that are created inner these teams of the provision group.

These are the relationships we have.
In that location is a data store on which we unsexed volumes. Directly there are story containers where we rationalise volumes in virtual machines. Regarding the fact that where it works, on that point was a slide, we volition not stop yet. I think that in the next releases there volition be an extension of the functionality, because now, in all likelihood, it doesn't even make sensory faculty to break up finely what on the dot plant and what does not work with practical volumes.

Well, something like this turns out like this, snapshots are pendent, replication, deduplication are supported, cryptography is supported and it turns impermissible much more flexible now the management of backbreaking disks. Migration, once more, hind end now be done, which possess non so far been done, only I hope that they wish provide it soon. So there they are, containers.

Here's a quick summary of what's interesting in the sixth Vmware. Well, in more detail, thither is, again, a ii-day course , with all the already detailed characteristics, and there you bum see, experimentation, ask close to.

OpenSource Virtualization

Another virtualization option that has been gaining quite an bit of popularity lately (although this virtualization is completely open, information technology is also supported by quite large companies). What kind of companies are these? RedHat brought it to nou, HP actively uses solutions from some RedHat and SUSE. This is Linux virtualization. Linux virtualization is supported in various shipway. Today these are two main ways.

The first is virtual machines - KVM. The second is the container. What are virtual machines good for? Virtual machines are the same virtualization as Vmware or Hyper-V. You create an operating system of rules, completely, it is altogether isolated from other systems, virtual switches, virtual storages, the functionality is around the same as that of Vmware. By the way, it works on the identical rationale as QEMU. Vmware remade it well, but American Samoa it were, its interesting features also appeared here.

KVM virtualization is a stable, good solution. There is even a kit from RedHatA named RedHat Enterprise Virtualization. They tried to go far look like-minded a V-center, only a little with its own characteristics. There is an open analogue, which is fairly primitive, to be guileless, I would not recommend using it in production, but for testing you can find out it, play with information technology.

But on that point is RedHat Enterprise virtualization that you can use to configure fast reliable efficient, and there is an open answer called oVirt. OVirt now, by the bye, is amalgamated aside OpenStack and if they complete the integration, we will get another version of a ready-made good stable cloud. If not completed, then we will use what is available today.

But also the KVM classic, there is another option that is rarely victimised, although in fact it is very effective in terms of resource management. This is called containers. Containers work not only in Linux. Here is one and only of these classic container options - Solaris zones. The example is exactly the same. What is the plus of the container? If you use VMware, you create 10 independent operating systems connected 10 virtual machines. Each of these operating systems uses the same kernel, the same libraries, uses disk space for these libraries and RAM to work. Yes, you will say that now virtualization, information technology has exhausted ahead, uses catching the same blocks of memory, a link to one block of sincere storage, saving resources, but in fact, ten virtual machines will always take up more than than unrivaled blank. Containers are one virtual machine. But at the kernel driver level, you create a twelve or much completely independent resources. Each of these resources has its own user area, its own mental process area, its own sphere of ​​disk systems, and only the main libraries they link to the homophonic area on the record. The lxc container ill-used to be. It has been used for quite an some time, but today, as IT were, it was replaced aside such an interesting attack called dock worker. Actually for what I started Linux virtualization. The lxc container exploited to be. It has been used for quite some time, but now, as it were, it was replaced away such an interesting approach called docker. Actually for what I started Linux virtualization. The lxc container used to be. It has been used for quite some time, but now, as it were, it was replaced by such an interesting approach called docker. In reality for what I started Linux virtualization.

What is docker?

Docker is a new container option. It was improved under Linux, but in fact it works subordinate otherwise operating systems. The difference between docker and standard lxc is that you forthwith create images of operating systems with ready-made applications. Plus registration in the cloud, creating I repositories, updating, deploying, deleting, energizing, deactivating - totally this is included in the finished halt again. That is, in fact, you get - a light cloud at the level of the operating system of rules, without additional funds.

What does it include? A node that ass create stubs, campaign applications, surgery push the repository. Hosts are our virtualization machines. Here are the containers into which we upload images, one, two, three OR more. And a secretary, from where we can take all these images or where we can give these images. In point of fact, you get both the cloud and the ability to work together along different applications. Yes, the drawback of much a system is that we have one OS, so you won't be able-bodied to emulate Windows on Linux. But you can employ the same kvm within lumper and do full virtualization for some other operating system. Nowadays longshoreman still works under the Mac OS. There are attempts to do something under Windows, you require to look ...

Question from the audience:"Does it work along Linux, or single on RedHat?"

At all. This is not a RedHat solution. It's right that RedHat immediately offers you to install altogether this into your repository. That is, as a matter of fact, you rump use any option. And there are options now how these applications can be migrated from one type of OS to other. That is, if you lead your configs, and altogether change all the binaries of the in operation system, you, in principle, can run along something else. Well, it's still not production. This is non a stable solution. That is, the problem is precisely that all this is in one operating system and you save both resources and magnetic disc space. But the system is completely isolated for assure.

Directly we pass to stage number cardinal: how to manage all this?

That is, what is the trouble? You complicate your structure, and at some manoeuver information technology turns out that your monitoring organization, already managing and monitoring, they get very colonial.

On that point are two options. To complicate the control organization further, Beaver State try to spread it altogether into several levels. Here is one of the management options. For a very foresightful time, HP worked with such a product, which was named Arrangement Insight Manager - SIM. Then he went into a large package called Inside Control. The package provided monitoring, installation, conversion, data collection, various automation options, etc.

What's close? Yes, aside the way, the package seemed to cost gaining weight, over the course of several years it grew from four gig to sixteen - this minimum of RAM requisite to act, yes, and then, equally information technology were, the moment came when it became essential to plug in sort o large volumes of servers especially all this was practical to the baskets, that is, the very blades that we have already talked about.

What did the blade let in? The sword enclosed and includes 8–16 servers, adequate 8 switches, a practical connexion appeared that allowed you to make profiles and transfer these profiles from one port to other, it allowed you to set macintosh addresses, vevens, cut the bandwidth - all this by small degrees it was introduced into insider control, although in parallel it was - virtual connect management. And so the developers decided to do this: and let's make a single utility that will include Inside Control and Practical Unite Manager and let us to work normally with this equipment.

The main tasks that were put forward: characteristic the part, collecting condition information, see how the various components are interconnected, incompatible systems for tracking events, how much work and work distinguishable tasks you are already doing, and of course, information that and how it works for you.

One eyeshot

What gives One View? Adult dashboard. You have already seen him. It shows you full information about your center, which includes servers, regular torture servers, which include blades. What's next? Depending on what equipment it is, you lav either draw full access operating room partial approach. Well, again, the certify gives either full access surgery partial direction access code. The disadvantage, by the bye, for OneView today is that the equipment that is older is not supported and most likely wish not be supported. Well, for deterrent example, the 3000th baskets are not substantiated aside them, and the 7000th ones are braced by them.

Further. Search is in progress. Card game. Map purview, which show you (we will see in a slightly larger format) various templates for elaboration tasks. Information for creating alerts about what works for you and how.

Configuration Examples

What are you creating? You have a server, you take up storehouse, you take up a trade. For instance, you take create a switch. You see: v-lan is sol-and-so, the band is so-and-so, attached or victimization networks such-and-such. Who worked with virtual connection, everything is real similar. From in that location and taken. These all points are taken almost completely.

Visibility configuration for the server: make a visibility, bind it to the chopine, group, set various addresses, vevens, poppy addresses, ports, bands.

Visualisation - the network, ports that are socially connected here.

Basket. Hera IT is - communication, components, all the servers that are enclosed, then again we again look at the one-on-one components.

The server is what we enter, how many blades, if we have several blades, so connected.

This is the approach we crapper already work with. About the resources. More than insider control. Of course, where without it. Information technology is believed that our retentivity is threepenny, then you can set up a draw.

Cloud substructure with HP Helion OpenStack

The next stage. You experience the base. You bear, let's say there, a bunch of esixes or hosts with RedHats or something other. Forthwith you want to either dedicate this infrastructure to the management of divers departments of your companionship or betray it. What chore? You need a tool. A tool with which a local admin terminate create 1-2-3-4-10-20 virtual machines, make up a switch, activate these machines, turn them off, transfer them to another host automatically, if something suddenly happens to you. For this, various cloud infrastructure options are secondhand.

One of the very intriguing solutions was and remains OpenStack. Well OpenStack as available origin software, information technology stiff pretty crude. Yes, this solution is good, but the job is that it requires quite serious fourth dimension, effort and knowledge to in good order configure IT. Various companies, large companies, practice OpenStack for their personal purposes. Information technology is powerful. One example is the HP Helion. They call it HP Helion OpenStack. What has HP done? They took the OpenStack root, their equipment and developed a system that will beryllium quite an unchaste and half-witted to install this very OpenStack on the right computer hardware. Well, how could it be differently.

Who started? Rackspace. Rackspace, National Aeronautics and Space Administration began to go, today Rackspace remains 1 of the near much immense developers, but others have appeared. Today, OpenStack is no longer even off a program. This is a frame, this framework includes a bunch of developers, a crowd of projects, there are standards that someone implements. Here in parentheses I wrote in descending ordination of the number of updates that are added to the general set of paper. That is, now HP came out connected top with a rather wide margin. Further RedHat, IBM, Huawei, VMware, Suse, Intel. By the way, VMware also actively uses it, and, among some other things, is developing its cloud body structure.

What does OpenStack itself allow? A few points, we will non dwell on them much. First, the creation of practical machines. Make, solve, update, cancel, that is, the entire life cycle. Today it is supported if it is HP, either KVM or ESXi. That is, either RedHat Suse, Beaver State it's VMware.
Separately - access to cube devices. HP either uses 3PAR or provides proliants with a multitude of disks. Thither are two options, and united option, and other option is available. Further, the use of containers, where objects are created, for case, create an operating arrangement, make up an lotion, turn it on, round IT off, delimit resources. This very Swift is called a container for management.
Full network access. Both switches and routers, firewalls, and freight balancing are included.

What else? Authentication system, individually. Separately, on that point is a arrangement for creating ready-made images of virtual machines. Dashboard for management, again, is a separate infrastructure, it may not be. Visualise management is backup, storage, and so on. Hardware interoperability services as symptomless.

Wherefore did I write out these very blocks? Because if you take the large deployment scheme of Helion, then it will follow immediately indicated: "The machine is much and so much, table service, swift, etc.". That is, if you take the circuit, you will immediately watch by these blocks what you will have on a finicky machine.

There is a deployment under KVM, there is a deployment nether ESXi, I did not find a classical version of either one and the other. Perhaps IT is somewhere in support, but they recommend either one or the other.

This is how IT is included in the HP Helion itself, here are our blocks that are present on individual components.

HP Helion Rack

Well, it's also a ready-to-wear computer hardware add called HP Helion Squeeze. IT includes a network, flex, the cabinet itself; support, again package, computing nodes, from four servers, DL data repositing systems is indicated, but you can use 3PAR, and one unit for control. This is a turnkey root, which is sold as an integrated complex. True, nobelium one bothers you to put together it yourself.

Question from the audience: "Four servers for computing and basketball team servers for management?"

Information technology starts from four ...

"I sentiment the fifth server for management"

When is the virtualization of IT infrastructure needed?

The system is that the cloud makes sense when there are a good deal of cars. You are for this tractability and isolation of the hardware platform from the user, you pay with ascendancy. If you ingest decade cars, to embody honest, you don't have to exercise the cloud. But if you lead off to grow, and then these cinque cars, they are already going at a normal percentage.

Let's just say that if you have a small cloud, you give the sack just use OpenStack, then these five machines turn happening as virtual machines, they work smashing both in lxc and in docker, by the fashio, this is exactly what whole kit for us. That is, we are building on lxc, and we already use this musical mode of operation, but if you take a ready-made system and this system does not have ten servers, it makes sense to take a ready-ready-made resolution. IT wish be faster and cheaper. Well, let's say, verified.

In fact, yes, if you cause a good company of developers, you can takings the same OpenStack and make it every last much cheaper, and pay the developers for the work. Well, you derriere not hire developers and straightaway use support. Practice shows that if there are atomic number 102 developers, information technology testament be cheaper. And quicker - 100%. Hither.

So shortly, I hot to talk about virtualization in my talk. Shortly why? Still, time is running out. You can tell a lot, Hera we have only 15 elemental courses in different directions, and then dig and dig foster.

Thanks for attention!

---

Report video

---

Distribution of Hewlett-Packard solutions in Ukraine , Georgia , Tajikistan , the CIS countries
Distribution of VMware solutions in Ukraine , Belarus , the CIS countries

Training courses on Hewlett-Packard
Training courses on VMware

Nearest courses in the Ml MUK (Kiev):
July 30 - 31 - VS6. 0-WN VMware vSphere: What's Recent [V5.5 to V6]
August 17-21 - VMware vSphere 5.5: Hurrying Track
August 25-26, 2015 (remotely) VS6.0-WN VMware vSphere: What's New [V5.5 to V6 ]

MUK-Service - all types of IT fixing: warranty, non-warranty repair, cut-rate sale of spare parts, contract service

DOWNLOAD HERE

GET Virtualization on Open source: report on new products and approaches to virtualization of corporate infrastructure / MUK Blog / Sudo Null IT News FREE

Posted by: gillianflery1969.blogspot.com

Share this post