CIO talks about how to extend data Center service life

This year is the 10 anniversary of the 1200-square-foot data center at Franklin Engineering College. This means that the facility is running longer than the planned time of the CIO Joannie Cossut (Joanne Kossuth), the vice president of operations, for a period of three years. Now, although the school needs a larger capacity and better connectivity facilities, Kossuth has been forced to postpone it because of uncertain economic problems.

In the past few years, demand has certainly increased, prompting data centers to reach their limits, Kossuth said, but the recession has shelved the renovation debate.

Like the CIO at Citigroup and Marriott International, Kossuth must be creative and get more from servers, storage and facilities themselves. To do this, she must re-examine the lifecycle of data and applications, storage array deployments, rack architectures, server utilization, idle devices, and other devices.

Lakes Coumars Rakesh Kumar, a Gartner research vice president, says large organisations have been constantly asking him how to avoid the cost of upgrading, expanding, or relocating data centers. Any data center investment will cost at least tens of millions of dollars, or even hundreds of millions of dollars. The average data center update rate is 5-10 years. That requires a lot of money. So companies are looking for alternatives.

While the prospect may seem bleak, Kumar has found that many companies have extended the data center's service life by 2-5 years by using multiple strategies. These strategies include consolidating and rationalizing hardware and software applications, and applying virtualization and physical relocation IT equipment. Most companies do not optimize their data center components, so faster usage limits are achieved.

Here are some strategies by it leaders and other experts to further extend the service life of the data center.

1. Migrate unimportant data. The first problem that causes the Kossuth of the Fu Ranklin Engineering School is the cost of processing data. For example, graduates, admissions officers and other departments take high-resolution photos of a large number of CD-ROMs per event. They use server, storage, and bandwidth resources to edit, share, and retain these large size photos for a long time.

To keep the data center from processing the nearly 10TB of data needed for these photos, Kossuth registered an enterprise account on the Flickr site and migrated the entire management process of the photos there. This not only saves her the 40,000 dollar cost of purchasing a storage array, but also eases the pressure on resource-intensive activities of data centers to handle high-resolution images.

Kossuth says the risk of moving non-core data out of data centers is small. Our storage space is now used for mission-critical projects.

2. Eliminate the pressure on high-priced applications and infrastructure. Earlier, the Orin Institute of Engineering purchased a 80,000-dollar Tandberg video conferencing system and supported storage arrays. Kossuth didn't run out of the investment because of excessive use, and now she prioritized video capture and release, shifting low-priority projects to inexpensive video conferencing solutions and YouTube for storage.

For example, most PR videos are generated outside the Tandberg system and posted on the school's YouTube channel. Kossuth says the data center no longer provides dedicated bandwidth for delivering video and provides dedicated hardware for video retention. More importantly, the Tandberg system retains much-anticipated record of meetings and remote learning of important tasks.

3. Standardize servers and storage. Dan Blanchard, Dan Blanchard's vice-president for corporate operations, said his main data center was 22 years old and he wanted to use the data Center for 20. He points out that it is the reason for the long life of the data center, especially in terms of standardization.

Every year the IT team chooses to buy some servers and storage devices. If you start a new project or if 300 to 400 physical servers are unavailable, these standby machines are ready to use. The storage will also be processed at the same time.

Even switches are standardized, even though the switch has a longer replacement cycle. Blanchard says consistency makes it easier to manage resources and predict capacity. If you have many unique hardware from different vendors, it is difficult to plan the data center.

Blanchard recommends working closely with vendors to understand the manufacturer's roadmap and implement phased standardization upgrades accordingly. For example, if a manufacturer publishes a feature set that is worth waiting for, Marriott International may delay the renewal of its plans.

4. Virtualize the devices you can virtualize. Blanchard is a virtualization enthusiast. Marriott International's physical server pool supports nearly 1000 virtual servers, freeing up ground space, power and cooling equipment. While virtualization requires high-performance and high-density servers, the fact is that Marriott International can consolidate hundreds of of energy-efficient physical servers to save the entire datacenter.

Gartner analyst Kumar agrees that consolidation is a positive measure in the data center, because some servers are used only for one or two applications, and this low utilization power supply is the same as the power consumed by a highly utilized server. Keep the server from 50% to 60% of the power supply when it is running at full capacity.

In addition, older servers are far less energy efficient than the newer servers currently being used to implement virtualization.

5. Identify and eliminate unused or low utilization equipment and applications. By taking the previous view further, Kumar said, the agency should "clean up the room". Data center administrators should use asset management software or other software tools to conduct audits to get an in-depth understanding of the application and hardware inventory.

What you like to happen is: According to Kumar, about 5% to 10% of the hardware devices either shut down or keep only a few apps that are rarely used. Back to the fact that the server consumes energy or other resources in any state of use, you can eliminate the application or virtualize it and eliminate or reuse the hardware.

' You have to make sure that every device in your data center is doing production work, ' says Kumar.

The international financial giant Citigroup's 14 data centers, the oldest of which used 20 years, regularly identify applications that are rarely used. Jack Glass, head of planning at Citigroup's data center, Jack Grass that something has been lost in the past few years. Not only do we use asset management tools, but we also patrol the data centers to ensure that each device is useful. However, he warns that the application team should be consulted before unplugging any of the device's power supplies.

Glass agrees with this view. If a low utilization application consumes hardware resources, the application should be virtualized. ' Virtualization must be our standard here, ' he said. If you can't cancel it, then you need to virtualize it where possible.

When the device is suspected to be in trouble, withdraw the device.

If there are other failures, don't worry about pulling the server, software, or storage device out of the field if you are experiencing data center capacity problems. Choosing to eliminate vulnerabilities gives you ample leeway to use your current data center for a few more years. Here are some options:

Choose to use the cloud service. Companies such as Amazon, HP, IBM and Oracle provide cloud-based services that increase computing capacity when needed. For example, if you have a seasonal project and this project will allow your internal data center to reach the infrastructure limit, you can temporarily order server resources from your service provider. This will meet your short-term needs without the need for long-term investment.

Using a portable data center. Using a portable data center is a solution if building or relocating data centers is not a concern for your organization. Manufacturers such as Cisco, Hewlett-Packard and Silicon Graphics Analysys provide portable data centers. These well-equipped portable data centers can increase your overall capacity without investing in large-scale construction projects.

Hire a venue. Renting a site is not something new in the Data center field. However, when you avoid renovating existing venues, renting a space can be your best friend. By migrating some of your infrastructure to the operator's site, you can free up valuable ground space, power supplies and cooling equipment. However, you can still maintain control over your hardware and software.

6. Control random testing and development. It leaders agree that most testers and developers will consume all the resources you provide them, and therefore must be carefully supervised. In fact, Citigroup's Glass says he often makes use of the utilization review often found abandoned or completed testing and development projects. Kossuth utilizes a forward-looking strategy to ensure that testing and development efforts do not exceed the load on her data center. She uses servers and storage resources as a sandbox to surround an area of the datacenter. usage is closely monitored. When the project ends, the resource is immediately absorbed. She calls this approach a way to protect data centers without impeding technological innovation.

7. Identify duplicate data in the storage array. Jason Kutticherry, vice president of data center planning at Citigroup, said the company took concerted action to reduce storage footprint, save data center floor space, power and cooling. The key technology for this purpose is to eliminate duplicate data. This technique focuses on duplicate data in a file and, in some cases, on the data in the file.

As a financial institution, we store a lot of data, Kutticherry said. Therefore, we want to make sure that we do not store multiple copies to increase the burden. Using de-duplication methods helps the company recover storage space quickly enough to avoid unnecessary deployment of storage arrays.

8. Promote the use of efficient code by developers. Kutticherry insists that while this advice seems to have nothing to do with the longevity of data center administrators, inadequate code can have a significant impact on datacenter resources. Poorly coded applications can force the server to work harder, consume more processing power, and increase the number of servers needed. Make sure your developers use the strictest code possible. In this way, the application is most efficient.

Data center administrators should also require developers to use a unified database rather than a customized database. Again, it facilitates a more optimized computing environment and reduces hardware and software limitations.

9. Readjust position. If ground space is your problem, consider relocating IT equipment. What happens in most data centers, Gartner says, is that once built and started working according to a certain design specification, new equipment will be added to the requirements of cabling and cooling in the years ahead, rather than optimizing the layout of the ground.

Kumar urged it teams to assess the ground layout every three to five years, and to use tools such as computational fluid dynamics (CFD) tools to create a suitable airflow model, or to redraw the floor plan using tools provided by the device manufacturer. Although the CfD tool costs about 30,000 dollars is somewhat expensive, it is worthwhile to extend the service life of your data center for a few years.

Kumar also points to a layered approach. This is a design technique for organizing data centers in an optimized way, with different parts evolving at different speeds or stages. So, in other words, Web servers are managed differently from 1-tier storage.

10. Analysis of energy efficiency. The CDF tool is also a useful tool for energy efficiency, in addition to diagrams that help you draw ground space. Companies often use a place in the data center to put too many devices in order to produce a hot spot where power and cooling are reaching their limits, Kumar said. By analyzing the temperature of the data center, you can postpone the purchase of expensive air-conditioning and postpone the increase of power. Simply rearranging racks or devices can add years to your data center.

Kossuth designed her rack so that there was a lot of airflow behind the rack, which allowed the device to cool properly. ' Our entire room has heat and humidity sensors, and if the temperature exceeds a certain range, it will be alerted, ' said Kossuth. This helps us to maintain optimized power and cooling for many years to come.