Analysis and Comparison of TCO in virtualization solutions

ViewTCOAnalysis and comparison in virtualization solutions

Total cost of ownership (TCO) is a commonly used technical evaluation standard, it aims to analyze and compare the total costs, including the first purchase cost (total cost of acquisition) and annual O & M cost, within a certain period of time. In some cases, this overall cost is an average of costs within the 3-5 year life cycle for comparable current expenses.

 

The comparison of TCO should clarify the premise that the implementation functions, performance, and reliability of the IT system are basically the same, or the degree to meet user needs is roughly the same. If we unilaterally pursue TCO and ignore the final performance of the IT system, we will put the cart above the horse. The best TCO system is not necessarily the best ROI system. Just as when we went to the mall to buy a pair of shoes, we probably didn't choose the one with the lowest price. The best TCO solution may not be the most suitable for us.

 

In this case, why do we need to focus on TCO instead of directly looking at the ROI? This is because TCO is the basis for ROI analysis. First, you need to understand I (investment in investment), and then analyze the expected R (return) to get the ROI. R (return) analysis is beyond the scope of IT technology in actual projects. For example, if a ministry wants to launch a new project a few years ago, the hardware selection should be considered as "not worse than a system just launched by the Customs ". This comparison is also within the r range. Of course, in the report, we will use the so-called advanced nature of the project to explain this return. Such as "leading in China", "filling in gaps in China", and "leading in the world. I once heard people say, "It is not difficult to fill the gaps in China because it is in China... Blank ". In short, the R (return) Part depends on the user's management layer. CEOs/CIOs/CFO/CTO think that the return is high and worth doing so, so there is a lot of hope for the project. However, it consultants and vendors trying to recommend software and hardware products should focus on the TCO of solution. After all, if you really have the ability to influence the decision-making of the user's senior management, these can be omitted.

 

In the current IT market environment, there is an interesting phenomenon. In the big talk, TCO is not a solution that seems to be relatively low-cost, but a solution that is "too high. So do these seemingly expensive solutions have a lower TCO? Or, how do they calculate the TCO to a lower level? Next, we will analyze the most common virtualization solutions.

 

When talking about virtualization solutions, we often hear a similar description: "virtualization and system integration can help users reduce TCO ." It seems that virtualization is used to integrate multiple IT application systems for management, reducing TCO. However, it should be noted that "TCO can be reduced" rather than "TCO must be reduced ". In fact, virtualization and integration may not necessarily reduce TCO. In other words, there are still many factors to consider.

 

Let's take a look at the reasons or factors that virtualization and integration reduce TCO. There are four main aspects: 1) improve resource utilization through integration, and reduce requirements for overall hardware configuration; 2) reduce energy consumption; 3) reduce data center requirements; 4) Reduce O & M management costs. Next we will analyze them one by one.

 

First, we will discuss "improving resource utilization through integration and reducing the need for overall hardware configuration. According to a survey, the average system usage of non-virtualized systems is between 5% and 20%, while that of fully virtualized systems can reach 60%. I do not doubt this statement because the application system needs to fluctuate. In order to meet the normal response needs during peak hours, the system usage must be very low when idle or low load occurs. However, it is very difficult to achieve an average usage of 60% after virtualization integration. I have seen dozens of international applications run on the same server. The so-called "Internationalization" means that users who use these applications are in different time zones in different hemisphere, that is, they go to work during the day and we are sleeping at night. At the same time, we should note that the above mentioned utilization mainly refers to the CPU utilization. Io virtualization integration is more complex. Both npiv and VLAN sharing technologies require good planning and complex mechanisms to avoid performance interference. In terms of memory, it is even more difficult to solve, because the application and the release of memory are not as dynamic as the use of CPU time, it is difficult to use the dynamic sharing technology of memory in real application scenarios. That is to say, the memory still needs to be so much non-virtualized, or even higher, because there is overhead (management overhead ). Even if dozens of applications share a server, the server must be highly configured. We know that the price difference between high-end and low-end servers is huge. The price of a server with 128 cores/512 threads is definitely not 16 times that of 8 core/32 threads. Its maintenance and warranty costs are also higher and not flexible. To sum up, "fully virtualized System utilization can reach 60%" is a condition, and there is a big gap with the current situation of our real users, which can only be said to be an ideal. The cost reduction in this part is not that easy.

 

Let's talk about "reducing energy consumption. Undoubtedly, new integration solutions can greatly reduce the energy consumption requirements of the original system. However, we should note that this is compared with the original legacy system. We all know that the performance of computers has developed rapidly, and the performance has almost doubled every other year and a half. Therefore, the energy consumption ratio (performance/power consumption) will also increase significantly. It is normal for new systems to save energy than legacy systems. The contribution of virtualization depends on the situation. This is closely related to the previous one. Only with good planning and improved utilization can we save energy. Energy conservation is very important, especially when we pay more and more attention to green. The power consumption of some large data centers is amazing, and the daily electricity cost is more than yuan. Some enterprises may not pay much attention to it, especially the it O & M department. CFO may be more concerned about this issue (there are too many things to be concerned about, so gradually improve it ).

 

I think virtualization will help to reduce the number of machines in the IDC room. A 32core server and a 4core server both occupy 4u space. The data center is also very expensive. If the original data center is full, you need to create a new data center. The construction cost is normal for tens of millions, not including the cost of housing and construction. Of course, even if there are few applications that require an individual "high-end" Data Center, this is another case.

 

Finally, let's take a look at "Reducing O & M management costs. This is where the cost is actually greatly reduced. I have studied IBM's scon and other system integration tools. The biggest part of the estimation is the cost of operation management. This is not hard to understand, because each IT application system was originally managed separately, and after virtualization, unified management of the Management Platform became inevitable. There is no way to separate hardware. servers, networks, and SAN storage are integrated into a unified management platform. Permission integration will inevitably adapt to system integration, and personnel will be streamlined. The management fee calculated by IBM scon and other tools varies greatly. After virtualization, the TCO is better. But does this apply to Chinese users? Most of our users are not easily considering cutting people. "Selling without making money (or earning less)" is a "Business is business." method of thinking that is easy to understand, but most Chinese users will not do that purely. Just as many people in the IT industry do not understand how IBM sold ThinkPad that year, but now it sells x86 servers.

 

We also need to remind you that virtualization will bring additional costs. Whether you are using paid management software such as VMware or an open-source management platform, there will be costs for use or development. In addition, this cost will increase as users use it. Even if users have powerful open-source and O & M capabilities, their own labor costs will inevitably be huge.

 

If you read this, you may think that I am an opponent of virtualization technology. No, in fact, I am fully optimistic about and support the development of virtualization. I don't think any technology can just say that advantages do not mention limitations. There are too many advantages of virtualization. It is said by promoters of various manufacturers every day, daily or yearly. I have also said a lot. Here, let's talk about some limitations to balance my previous comments. After all, balancing is lasting. I hope everyone can make better use of virtualization.

 

 

This article is from the "rong Yun infrastructure optimization" blog, please be sure to keep this source http://rongxin.blog.51cto.com/8955502/1429860