CONVERTING A DATACENTER TO BLADE TECHNOLOGY

Benefits And Disadvantages Of Converting A Datacenter To Blade Technology

Converting A Datacenter To Blade Technology

Introduction

Bladed servers are all rage in IT departments around world. Manufacturers have developed the whole line of fantastic bladed server devices designed for service provider and high-density enterprise applications. However, it is important to understand that these systems have both advantages and disadvantages that must be considered before IT managers place an order.

Advantages And Disadvantages Of Upgrading Server Hardware To Blade Server Technology

The demand for these systems is based on manageability, ease of deployment and efficiency of space utilisation. All three of these needs are met with bladed systems. The console management and remote power control of these systems are powerful, and getting better. Corporations managing hundreds or thousands of servers enjoy consistency and ease of these new bladed deployments. Companies with rapid growth, in particular Internet companies, have found that bladed systems deploy faster and with less installation troubles than alternatives. At Equinix, where eight of top 10 Web properties in US host their gear, there is an increasing number of deployments of these systems, but they tend to be used by less Internet-centric organisations. In any modern data centre you will see various examples of these bladed systems, though most are still experimental and for smaller applications at this time(Farley, 2004).

Part of slower adoption of these high-density systems are because bladed solutions are not for everyone and every application. There are the number of reasons why bladed servers might not be best solution for some organisations.

For example, one large Internet search engine that considered deploying bladed servers chose the large-scale 1U server deployment instead. For this company, added advantage of having individual devices, where the single device failure would not take down an entire bank of servers, was critical. With bladed server systems, multiple CPUs rely on the single device, increasing risk of potential failure. In this case, manageability is exactly same as bladed servers - company uses identical 1U boxes with common management, console and power cycling tools. However, servers are much less expensive per CPU, and in theory the single server can disappear entirely without affecting entire system. A number of financial institutions have taken same position where deployment is concerned. While rolling out blades is in some ways easier than the high-density rack of 1U servers, failure management is easier to modularise with 1Us. When determining the cost per CPU deployment, you also must consider replacement costs in staff hours, shipment and equipment. A single blade can cost significantly more than the 1U CPU box. In high-density deployments, cost per 1U server is so inexpensive that we have seen these financial institutions treat 1U boxes as disposable devices. They remain "disposed of" until next scheduled staff visit to lights-out data centre. Finally, the common misperception about bladed servers has to do with space efficiency. While it is true that you can fit more individual CPUs per cabinet using bladed servers, power consumption drastically increases per ...