Talk about clouds and a review of other interesting reports from the conference Datacenters Innovation Forum 2016

SIM-Networks Technical Director Yury Bonar told about different conceptual approaches to cloud solutions and gave a brief overview of the difficulties customers face when planning to migrate to the cloud, the infrastructure architecture for their services and projects.

Last Friday, September 2, a conference for data centers took place in Kiev, 90% of the attention was paid to the clouds and everything connected with them. The event was quite interesting, because not vendors (there was only one - HPE) shared their experience in implementing and using different technologies for clouds, but integrator companies that “twist the nuts” and cloud service providers. I am quite a lot at all sorts of excellent conferences, and I must say that this one was rich, all the speakers spoke on the matter, and based on their practice, and not only on theory. Under the cat review of the event, as well as theses of the reports on the clouds, for example, it was about such an important issue: how to choose a cloud provider and various nuances on the clouds.

')
Classical cloud deployment scenarios - when cloud-based solutions deploy their services and resources for the purpose of resale to external customers or to provide services for internal customers. As an example - startups - test sites for further growth. There are no capital costs here - they took, used, tested, further or expanded or removed those resources that are no longer needed.

Creating a backup site for the company. Those tasks that do not allow you to fully deploy in the clouds for some reason, or, most often, for security reasons, very often use cloud infrastructure as a platform for duplicating or backing up your data - to have hot copies.

Fight peak loads — you can dynamically expand or reduce on-demand / schedule computing power: days, hours, periods, etc.

Complete replacement of the physical infrastructure - yes, now, many people look skeptically at the clouds for various reasons, but everything is moving towards virtualization, and most likely in the future, service providers will be engaged in hardware.

What worries customers about choosing cloud solutions? What questions are most asked? Most often, customers are worried about the security and complexity of migration (readiness of the infrastructure for migration), how many simple services there will be, etc. Scalability and performance - as far as it will be available in the future, whether they have enough scale of these solutions in the future. And whether the cloud will work no worse than their physical server is now.

Difficulties in assessing financial performance. The issue of security when moving to the cloud is rather a matter of trust to the provider. The client himself must be concerned about their security and data backup - encryption of important data.
Fault tolerance and protection against DDoS attacks — by their structure, cloud solutions already carry the necessary elements for duplication and replication of data. There is a single point of failure already at the level of iron - the provider should have all this if the cloud provider approaches the cloud structure seriously and professionally. In respect of DDoS attacks, the cloud provider already has protection, or this protection is targeted when a client is attacked by a service.

The separation of data is dictated by the cloud options that the client chooses, this is achieved both by using the SLA and by using technology that alone guarantees a certain level of separation and security. The clouds in many ways simplified the life of the atishnik, who do not need to wrestle with and be constantly to some extent the architects of the network. Business only thinks about how much resources it needs - RAM, disk resources, etc. and not in those situations when the admin says, "but we cannot add memory, as the server space has run out."

Faster and easier miscalculation and predictability of expenses, bidirectional scaling of business processes. Improving resource allocation. Cloud technologies are a “trend in trend” in the movement to the Internet of things, when devices become more and more simple, they don’t have any power, everything is taken out of the limits of this device, into the cloud, or somewhere else.

The clouds include a whole set of market trends, this is a very wide current at the moment, many models of service (s). When considering the transition model from its virtualization to the cloud. As a rule, there are 2 options when you rent a cloud, or you have your own equipment - in this case, the load on sysadmins increases, they always think about where to remove the virtual cloud, where to remove resources, where to add, etc. Thus, technical specialists are not worried about what is needed for the business, but about the fact that there this iron can still be done - that is, the constant questions: did he think through in advance how much he could expand the possibilities? - this task of resource management rests on the task of competent configuration of iron. It is often possible that the equipment being purchased for 3-4 years may not be enough for the intended use, or it will not have the necessary functions in the future, and the current configuration of the equipment may not cope with the arisen tasks. Buying equipment for the test is not profitable, and you can try the cloud and not have an excess of costs.

Resiliency of different levels - the cloud provides it at the virtual machine level, but what happens inside the instance - the cloud cannot control it - if your service has run out of memory, it will look from the cloud that everything is working. Cloud solutions ensure that the virtual machine is up and running, and its parameters will be as stated. On the client side, there must still be monitoring of their services.
Questions to the speaker from the audience showed that it is still very hard to “break away from the iron.” And the listeners were also very disturbed by failures in the cloud. Nothing works without failures, there will always be failures, the question is how they are solved. In the cloud, clients often do not see failures, the maximum that can be is a complete bankruptcy of the provider, when a client once arriving does not reach his cloud. But this is a matter of choosing the provider and trust - which is lower.



The head of the SIM-Networks project department, Miroslav Migovich, talks about the nuances of choosing a cloud provider.

Many companies do not know where to start in relation to the clouds. They heard, but they have some barriers that do not allow this first step to be taken. We will be glad if these recommendations help someone.

What to consider when first approaching the clouds and How to choose a cloud provider?

Theses of the report:

The concept of clouds is blurry, if to speak very generally - this is a software and hardware complex, which is located on the site at the provider, and whose resources are leased. For the user, it doesn’t matter where the power is. The user's computer should only know the address at which it receives data. Ultimately, the user gets the same picture on the screen, as is the case with a regular desktop, and he is not worried about such things as cloudiness and virtualization. An example from our life: in the 90s, when users first started using e-mail, they had no idea where their letters were, on which continents and countries, they just used this resource, and at that time we didn’t even know what we now call this service is cloud.

From a business point of view, when moving to the clouds, the IT infrastructure of an enterprise can turn from a deterrent into a powerful catalyst for development. Of course, taking into account all the necessary factors. The dynamics of global economic processes in the world lately forces enterprises to choose the technology that will correspond to adaptability and flexibility due to the changing realities of the market, and it so happened that today it is cloud technologies that meet market demands. Looking at trends, we can say that the approach changes from physical solutions towards cloud solutions - and this is not a fad. If you look wider, when people stopped creating such resources as heat and food in their homes - now we turn to suppliers - we go to the stores (and in the LCDs, if there is no heat :).

How to get all the benefits of cloud technology?

This may seem obvious, but many providers forget about it. A very important feature of all cloud projects - each project is considered individually, the cloud is not a panacea for everyone. You might think that “everyone knows this anyway,” but often providers do not hear what the client says and do not want to see what this particular client wants. Think through a competent process of migration to the cloud. If necessary, add a testing phase, add all the necessary resources - thus helping the client to get acquainted and make friends with cloud technologies.

We all know about the concept of a public cloud and a private cloud. They are built on the same technology, on the same structure, but there are some important distinctive features. A private cloud is its main feature - it is isolated from other clients at the provider. This model is always recommended for those companies where data is particularly careful - financial databases, etc. In the private cloud, the client gets maximum control over their resources and maximum security.



In the public cloud, resources are used by many clients. Of course, cloud load balancers are involved here, which automatically allocate resources, which helps the provider to fully use their resources. Who uses public clouds? As a rule, these are separate divisions of enterprises, or individual employees, separate applications for divisions — which are not critical — they are secondary. Then the question is: why are public clouds so popular? They have obvious advantages - a very affordable price - it’s very quick and easy to deploy services in public clouds - you can very effectively convert capital costs into operational ones. Using public clouds, you can quickly regulate the pool of resources needed by different departments according to different criteria - season / non-season, time of day, season, etc., so that needs and capabilities match. This increases the efficiency and computing resources and financial.

Hybrid cloud Why are they, according to some sources, the most popular? They allow a very thoughtful and balanced distribution of business units — individual services, individual employees, and data on different types of clouds — depending on the criticality and relevance. Plus, this model allows you to better use financial resources - as a private cloud, whatever one may say, is more expensive, with all its advantages.

A serious lack of a public cloud - the degree of security and control is noticeably lower in comparison with a private cloud.

What do our customers pay most attention to when they start working in the cloud?

First of all, business pays attention to the unprecedented flexibility of this technology. And, interestingly, the very notion of flexibility used to be abstract, since it used to be difficult to calculate the correspondence of resources, business processes and capital investments. And the process was long and not very accurate. Now, having a pool of virtual resources - and using only a few numbers, it is possible to calculate the necessary resources not only up to the department, but also for each workplace.

Efficiency. As already mentioned, the customer does not need to know where the resources are located, he wants to know only that his infrastructure is functioning properly. Since the company pays for only those resources that are needed, thus achieving maximum efficiency. Investment attractiveness - there is always an emphasis on "cost of ownership" - the clouds significantly reduce the cost of ownership of infrastructure, according to some studies to tens of percent.

Controllability. This is one of the main priorities for many customers - the degree of consolidation of resources is maximum if the company has moved to the cloud. Infrastructure management is sometimes possible by one person from anywhere in the world. In addition, such an administrator will be able to easily recover not only data, but virtual machines - with the proper organization of archiving.

An interesting fact that is observed abroad, with the introduction of cloud technologies - increases the efficiency of labor. As resources and time are clearly limited, the efficiency of working time is growing.

As for reliability. This is an important point that we always talk about with our customers. At its core, the cloud is fundamentally different from the solutions we are used to. All nodes, all communication channels are duplicated, at least according to the n + 1 scheme. This allows customers to absolutely not worry about physical security and fault tolerance solutions. The client takes care of higher levels of abstraction - levels of OS, settings, applications, data. Everything else is already on the shoulders of the cloud provider.

Availability. This is especially for enterprises whose geographically branched structure - this technology allows you to consolidate data in one place - and all remote employees and branches can use the data centrally - and this is already embedded in the technology itself, and the company simply uses it.

Your data is the most valuable thing you have. You yourself must be concerned about their security - you need to encrypt the channels and data. This seems trivial, but many clients do not do this, we suggest what and how - then everything becomes good, but initially it is necessary not to miss this question.



It is always necessary to provide for the correct archiving of data. In the cloud, it is convenient to archive not only data, workstations, but also entire virtual machines. At the same time, you can use for this not only local storage facilities that are located in the same cloud, but also remote sites - which will make your infrastructure not only fault-tolerant, but also disaster-proof for minimal capital investments.

Free migration. Very often the company has the need to transfer data. It is recommended to choose those providers that offer free data transfer. There are such providers, you do not need to pay for each movement, you just need to find the right provider.

A few words about the tariffs. Unfortunately, there are still providers that charge and limit traffic. Obvious advice, look for those providers who offer unlimited tariffs.

Bandwidth inside the cloud - this item is very important from a technological point of view. This is a scrupulous moment, many suppliers are reluctant to talk about this topic. They do not want to reveal the structure of their network. But if you look at the essence of the issue, then it depends on how much quality service you buy from your provider. It depends a lot on the speed inside the cloud. How are the nodes connected? at what speeds? What protocols? Services of many enterprises may well work on 1 gigabit, but you need to consider that most providers use 10G, and only the most advanced build their clouds on 40G, or InfiniBand 56G. It is recommended to look for such providers in order to get advanced infrastructure in the cloud without additional investment.



The second difficult moment is the core. We all know that rented and virtual cores in the cloud can be very different - there are different generations of processors, different frequencies of buses, different memory speeds. Feel free to ask questions to providers - what kernels do they offer you, you should know what you pay for.

A few words about the migration process. If we roughly divide providers, that is, the part that offer their services "as they are" - take what we have, use it, but how we do not care. But it is better to look for that provider who will not only be your seller, but will be your partner and consultant you can rely on, and he will not leave you in trouble.

Some clients have branches whose applications are not supported in a virtual environment, so it’s better to work with those providers who can lend you physical equipment as well. This complexity - it can help you very much - you can get a full range of services from one point, which also saves money.

Security data centers. Too scrupulous moment, considering the various events in this area, we recommend choosing those providers who not only guarantee the safety of data, but also the physical safety of your equipment, which can protect from a certain situation in the country, and from some actions of regulatory bodies.

We can not ignore the fact that our legislation does not regulate the withdrawal of equipment and customer data from data centers. Severe violations and seizures have occurred, unfortunately, and this situation will continue until legislation is passed. When choosing a provider, also study the legal aspects of the country where the provider stores your data.

In our country, the concept of cloud computing is only beginning to penetrate into the minds of the business community. The obviousness of the potential of this technology makes many managers rethink the classic approach - this is a common phrase, but it is. Many need to change their IT outlook and destroy their stereotypes. There is a very important psychological factor on this path - many are simply afraid of this new technology. At the initial stage in this psychological plan, trust in the provider is the most important aspect. Part of the time in communicating with customers is dedicated to helping the customer “make friends” with the clouds and break stereotypes. After all, sooner or later we will have to use this technology - after all, we are already using cloud technologies everywhere, just don’t suspect it.

Next, fellow integrators talked about how they implement and test cloud solutions.



There was a report on security in the clouds. Since many vendors see this “cloudy” trend and on the one hand try to hear from the market any requests for security, on the other hand - they try to guess where and how security will be needed when using clouds. Of course, with all the advantages of cloud technologies, no matter how good they are - their use sets new security challenges.



The speaker talks about the VMware VSAN product, he said that most often this product is being introduced to develop and test solutions for time-limited projects. It is also important for service providers to increase the power of their data center linearly on the fly. On this technology some zody in Russia and Europe are built. Since vendors love OpenStack and are beginning to actively use it, VMware takes into account this trend and tries to keep up with it in this direction. VSAN technology allows not to grow new servers and use old ones, there is support for blade drives. The speaker shared a trick from his practice - if you don’t want to spend on SSD drives to test this technology, VSD can mark the HDD as SSD - of course, it will work slowly, but you can see how the technology works. In the new version, you can monitor each machine separately. VSAN is comparable to midrange skhd. VSAN licensed per processor (no cores and volume)



A colleague from Latvia spoke about the experience of implementing a VDI solution for large corporate clients in Latvia and Ukraine. Initially, there was a negative experience of implementation on Citrix (in the call center, the voice was transmitted with problems). Implemented the project on VMware, via Horizon 6, the delay was 40 ms, but VMware promises that VDI built on their solution will work with a delay of 200-300 ms. He said that the representatives of the client worked on the VDI on the plane - everything worked, of course, not as fast as usual, but it worked without failures. He told about the importance of debugging in VDI-projects. As in such projects, the implementation is both installation and debugging, that they installed something for 1 week, and tested the solution from 4 to 6 weeks - you cannot leave it without full debugging, otherwise the client will be dissatisfied. More abstracts: - there is no single VDI solution for all clients; - not all problems can be solved VDI - this is not a panacea; - anti-virus protection in VDI solutions is carried out in a special way: a special server anti-virus; - VDI - very convenient as a single backup for all.



The speaker talks about how software developers can use NetApp cloud solutions.
Cloud providers always say - the cloud never falls, but gently hints - it is better to keep archives somewhere. Example - Salesforce, when the cloud fell, they had a backup option of switching to another datacenter, but they lost 2 hours of their customers' data.



The speaker talks about the decision to optimize the channels when implementing cloud solutions. You always need to optimize the WAN channel, no matter how good it is, since even the bandwidth will be good, you need to plan the distribution of traffic, because without it any change in the volume and data request can cause complications in accessing cloud resources. It often happens that if there are problems with IT infrastructure, “networkers” - they say - this is not us, the programmers - our application works well, and this is not just shifting responsibility for failure, it can be really difficult to establish the facts. The speaker spoke about the decision that monitors the entire cloud infrastructure, and helps what is called to get to the bottom of the truth. It is much more profitable to programmatically build, distribute and administer your network. But, the program network requires more precise tuning in different variants of application of various solutions or their connections. Often, the client gradually becomes aware of and relies on cloud solutions - if the client first has only 1CB, the cloud has moved, then after a year there is already a mail server, etc. and even if these services on one cloud of one provider are not one infrastructure - the task is to combine all these parts into an already new, cloud infrastructure. The work of the integrator to calculate that the client with a new, already cloudy office, had no problems.



The rapporteur from the HPE, in his charismatic report, first spoke about common things, said that at the present time it is not the one who is the first to introduce some kind of technology who wins, but the one who is the first to introduce the very idea who will be the first to provide services and not only "will give technology." As an example, he cited Prozoro’s system for monitoring and implementing and auditing public procurement. As before, not only the technology was not implemented, but even the idea of ​​such a service was not implemented anywhere in Ukraine. He further talked about what HPE can offer to cloud providers and integrator companies in terms of supporting and implementing cloud solutions. For example, it was about a composable solution - HPE Synergy. HPE - does not replace cloud providers - it gives them more efficient use. HPE, as hardware platform makers, allows cloud service providers to better implement them. The introduction of a service is to sell outwards, or to provide an inward - in-house service, how quickly it is done - depends on the “clouds” themselves.
It also talked about infrastructure as a code — the tendency to maintain a modular, easily automated infrastructure, and to describe this infrastructure using a high-level language. The speaker announced HPE plans for equipment that can be used for clouds as well. He told about the vendor’s vision of the future of the IT industry, about updates, for example, on September 26, a new release of management software will be announced, up to 480 servers can be managed in one pool. He said that the HPE system already provides 240gigabit server interaction with the server, without narrow necks. In a word, HPE wants to overtake the future in terms of the capacity of its equipment, focusing on the speed of 600 gigabits in the near future.



There was a speaker, a CIO from the company, a large retail network KBT / MBT, with a report on how they transferred most of their infrastructure to the clouds, how the provider was chosen, how the project was evaluated and done, how the architecture was done. Moreover, the data and structure (s) are divided geographically. Why exactly the speaker was not told so :). The engineer has already told from his experience that it is impossible to predict all corporate and client traffic and how much the applications will be involved - all this grows nonlinearly. If earlier it was one CRM in their company, now it is already 20-30 applications. He cited examples where the burden on the IT department is greater, for example on Fridays — they always have sales and discounts, so IT on all Fridays orders more resources from providers. There is also an adjustment to the time of work of employees (remote employees) when people work - resources are involved, and at night resources are released for other tasks. This temporary increase in load can be either vertical or horizontal. , , . 16% , - . ( ), , , , , .


- SIM-Networks .

P.S. , – , – .

---

SIM-CLOUD - Fail-safe cloud in Germany

Dedicated servers in reliable data centers in Germany!
Any configuration, quick build and free installation