Cloud in the system administrator everyday work
Posted by on 19.02.2014 22:23
The first reaction of most administrators to the typical advertising of cloud computing can be quite careful. Companies receive a promise that the cloud will reduce costs and simplify the management of IT infrastructure and applications, and that can only mean one thing - layoffs. Of course it is not like that. No one and nothing can replace the administrator in your company. The cloud provider is not always able to take over its tasks, but it may help him in his work, giving him the tools through which it will be more efficient and more enjoyable.
Cloud solutions provider, Oktawave, indicates five primary areas in which IaaS (Infrastructure as a Service) cloud can help administrators in their everyday work, whether they work in small start-ups or large corporations.
Cloud as a sandbox
Sometimes you have to perform in a production environment some test operations and we know that this creates a certain risk. In many organizations running test environments on physical hardware is sometimes also a problem - it requires the consent of managers and signing many documents.
In contrast, an account in the cloud allows you to log into the control panel, run an instance of a virtual machine based on whether it is a pure system or a preconfigured environment, set a few options, login via SSH or RDP, see how it works and finally switch off the machine. If necessary, you can clone the machine, checking the effect of different configurations on their behaviour. Implementations that last many hours, in the cloud come down to minutes.
Moreover, you don't need control panel and all that clicking on the browser. Everything can be automated using scripts in Bash, Python or Perl, or with help of tools such as Git and Chef. One script allows you to manage the entire cluster of virtual machines: a well-documented syntax of the command line tool allows you to quickly forget about the browser and manually managing the virtual infrastructure.
Importantly, it is easier to work with virtual hardware than physical. Not enough memory for the database? Just increase the instance type. Too slow disk? One click and the drive accelerates to a level that is difficult to obtain on the physical hardware. Now imagine a situation in which we are trying to convince the IT director that we need to quickly buy more memory to the server. In the cloud it is not a problem.
Cloud as a shelter
Appropriate use of virtualized server, virtualized storage and virtualized database solves the problem of backup. Whether it's running the quick master/slave replication from the production environment of PostgreSQL on the local server or synchronize files between corporate NAS and drive in the cloud by rsync or for storing large video files in containers of cheap, object storage - everything can be deployed and configured using the familiar Linux tools, moving the responsibility for reliability and security of virtual infrastructure to the cloud provider. Identity management policies and access to resources in the cloud allow control over who, how and what can use.
Cloud as the balloon
By launching a new service or putting a new application on the server, it is difficult, at the stage of internal testing, to predict how it will behave when it is available to end users. Large organizations can afford many server clusters at the beginning, for smaller companies it is difficult to find the money for it.
Autoscaler working in the cloud solves this problem. No need to configure loadbalancer, because it works in the cloud automatically. Just start with a small instance, set the upper levels for scaling and asign it to different groups of virtual servers and watch as the increase of loads, turns on another machines and how their performance is increased - and how they return to previous levels when the web traffic fades.
Thanks to the cloud, also a small start-up is able to start a popular network service and not fail financially because of expenditures on equipment that does not necessarily have to payoff after all.
Cloud as a supervisor
Even if - for these and other reasons - organization is holding the server infrastructure under own roof, you will want to take advantage of independently hosted tools to control condition of infrastructure. How much are worth the control tools if they fail with your own infrastructure? Meanwhile, small instance in the cloud is enough to run a monitoring solution, which will supervise the local servers using tools such as ganglia or gkrellm or by sending a notice in case of any problems.
It is even easier if you completely move to the cloud infrastructure: external monitoring services for virtual instances are available and they allow to detect the problem from the inside, especially related to the availability of services to the end user.
Cloud as the Swiss Army Knife
Speed of running any services in the cloud, their reconfiguration and their ability to connect with each other makes the list of tools in the cloud endless. In the case of infrastructure, in some organizations, use of cloud makes sense only for the periodic operations - for example servers backup performed daily. In others, it may be that the cloud should run only scripts causing very high load on the server, as in the case of generating thumbnail images or PDF documents that - if running on the main machine - would harm the reliability of its services.
Cloud can also be useful when you need to run an application that requires a completely unknown environment (for example, when using mainly Java/Spring, for some reason you need to use a tool written for Rails) - then it can be treated as a black box in the entire system, without going into mode of action. Low cost of entry into such operations means that only the ingenuity of the administrator is the limit.
How cloud is changing the administrator job?
Cloud does not deprive the administrator of job, it only changes its character. Rather than manage individual machines, you can manage whole groups of virtual services that run as separate fragments of a virtually unlimited resource pool of computing power, memory and network. The limitations are only:
The speed of the cloud also has its other side - previously used administrative tools are insufficient, they simply do not scale with the growing virtual infrastructure. Earlier, you used a blend of scripts, proprietary tools from software manufacturers, manually configure systems and indestructible vi - now you can get lost.
Updating ten and hundreds or even thousands of servers is a completely different matter. You need tools and techniques to automate these processes, which will help connect all interdependent services to network at the same time. The administrator who uses the cloud must therefore be smart and be able to increase their productivity through the use of clever tools.
However, you cannot learn it all at once. Start by experimenting with cloud - get to know the services available and the differences between them, become familiar with tools such as Chef and Puppet, check mass services management. It is also good to be interested in the programming aspects of the clouds: knowledge of the API does not hurt, and can be very useful during the experiments and at the preparation of production systems.
The next step is to use the cloud for production use, once you know what types of instances are needed for which tasks and how to combine services in the cloud with what works under their own roof and there is no need to manually configure servers.
The last step is the ability to optimize the cost and performance of services in the cloud, transferring some of the burden from the cloud back to your own infrastructure, helping developers in problems arising from the start of applications in the cloud or dealing with licenses for commercial software (which in the virtualization field can be complex).
There is no need to cheat yourself: public clouds due to technical, legal and purely mundane reasons can not entirely replace the IT infrastructure in larger organizations. The future is a hybrid - and cloud-based solutions providers want system administrators to see in this a chance for themself and not threat to their position.