The number of construction in clouds that are doomed to failure is staggering. Which is a shame because manufacturers are getting better day by day.
Let me explain.
At Citrix, we get the benefit of talking to many people who are actively building clouds. They come and tell us about various projects they are working on; VDI networking for basic virtualization. For example, not too long ago, we talked with a team from a systems integrator (SI), which had plans to build their own cloud to offer their customers - not an uncommon strategy. After laying their plans for cloudify computing, storage and applications, I had to ask about their networking plans.
Like many others, they really had not thought of. The assumption was that "networking was already there" and no other thought was necessary. As you can imagine from the title of this blog, saying "the networking is done" in speaking of the cloud is a cloud to condemn to failure
Here's the problem :. Two of the main goals with clouds is multi-tenancy and elasticity. Networks, as they are traditionally built today do not realize one of these aims. Most networks are traditional hierarchies with several levels with a series of silos for application proxy. So when a single instance (either an application or tenant) must stretch beyond the hierarchy can support, bad things happen. Manual changes are required, assuming that the change can be supported at all.
For example, consider the simple case of having wide infrastructure with computing resources. The right way to do this is to have the network infrastructure to work with the level of orchestration so that the additional load is detected, additional servers are spun and the network is configured to automatically put these new online resources . Lose elasticity, lose value.
On the surface, this seems geekery. Except it is not . Build a network that has features such as multi-tenancy, elasticity and coordination with the orchestration requires a game plan is laid out in front and from the top. The IOC needs to work with its management team to ensure that all players are eligible to be part of the solution from the beginning. Operating from a single point of view and try to include other elements (eg, starting with calculation and then try to wrap it in storage and networking later) are safe ways fire to give a result unoptimized.
Fortunately, Citrix is busy working on tools to help.
There are not enough to build products. we really see the next step is to provide solutions that integrate different components as much as possible so that they work off-the-shelf. In the case of cloud and networking, there are three key aspects to the solution:
Making the Cloud Ubiquitous . Between CloudStack, XenServer and NetScaler, Citrix occupies large footprints in over 75% of the cloud market today. We know what makes them tick and how clouds make them succeed. But to really make their ubiquitous requires building an ecosystem of developers, tools and partners. To make the first move, Citrix recently contributed part of CloudStack to the Apache Foundation together to pull the juice from the developer community. This kickstart active tinkering so we can see the next generation of tools and products designed with the cloud as an integral component in mind. For publishers of traditional enterprise software, an Apache CloudStack provides a particularly interesting opportunity to connect in a truly open ecosystem where their work can remain theirs if they wish. We expect that this will lead to greater involvement of the major software vendors and a Starter growing community is good for everyone in the space clouds.
Integrating networks . CloudStack has integrated networking elements early on with various open source tools based on Linux providing routing batteries, NAT capabilities, and automation. But the way a router and firewall are considered key elements of any modern network, quality controllers of the service delivery business that offer features such as L4-7 load balancing become as much of a requirement if not more. NetScaler in particular, is a strong solution in this space and is a start with the work of active integration much like trivial integration API past. Developers are already asking how they can improve the workflow for corporate deployments and keep the promise of cloud around scalability, elasticity and multi-tenancy. We expect 2012 to be an exciting year on this front.
Automate . Forget how the network is a sure way to condemn a cloud, forgetting to automate is a similar question. Automation is particularly striking when looking at how the network is automated so that it can dynamically act and react to application changes, changes in rental, etc. Tools for automating the well documented API through are first on the list to ensure that both conformists scripts and software companies have what they need.
This is how we will ensure to network in the cloud. Now it's your turn .
top back down . Take a look at the strategic plans of the clouds that have been set out and be honest with yourself: Is it really up and down? Are all actors involved? They are a significant example? If the project started from a team or there is one team "cloud", then chances are that something becomes the tree in the design and it will be a problem later. Start top down and be honest -. Ensure that all stakeholders are at the table
Revisit your default . Every large organization, be it a company, telco, or SI had a set of default values to them when they need to achieve the common task done. Need another disk rack? Call EMC representative. Need a switch or router? Call Cisco. Even Citrix has been a benefactor in that need a remote desktop? Call Citrix. But when it comes to the cloud, all paris open. Each vendor has their cloud approached differently and all methods will not be consistent with your strategy. Revisit your default choice for the tools you use. This does not mean existing suppliers can not work. - This simply means that sellers even existing need an honest assessment to ensure they are compatible with your cloud strategy
Revisit your process. There are opportunities, operations team has a number of well-defined and well thought through process. While elsewhere in the office, someone asks why the great players of clouds become one administrator 4000 servers. Much of the answer comes down to the process. Too many processes are not automated well today and make a success cloud, not to mention the network, it is based on achievement, requires automation. APIs are the new CLI and processes that do not take a hard look at how automation can simplify the steps are doomed to failure
The net-net of all this chatter is simple .: the rules for everything we do in our infrastructure is changing. The things we have taken for granted, such as networking are completely redefined. The criticism most if key elements such as your network are not a part of your new underlying design, it is almost guaranteed that your big plans for cloud greatness fail.
I do not want augurs ill will of the infrastructure, but if you do not want to condemn your cloud project, do not forget networking.
0 Komentar