Takeaway: Going to the cloud might make a lot of sense for your organization — but only if you choose a secure, reliable data center that can handle your needs. Erik Eckel runs through 10 essential criteria for data center evaluation.
Everyone's going to the cloud. The cloud's all the rage. Almost no IT discussion is complete without mentioning "the cloud." But when it comes down to it, the cloud is nothing more than systems hosting information in a data center somewhere "out there."
Organizations have discovered the benefits of offloading infrastructure development, automatic failover engineering, and multiple coordinated power feeds, not to mention backups, OS maintenance, and physical security, to third-party data centers. That's why "going to the cloud" ultimately makes sense.
Unfortunately, not every data center is ready for prime time. Some have sprung up as part of a cloud-based land grab. Review these 10 factors to ensure that your organization's data center is up to the task.
Note: This article is also available as a PDF download.
1: Data capacity
Data centers are typically engineered to support mind-boggling data transmission capacities. Some feature multiple OCx and SONET connections that can manage Amazon.com-like Web site demands. Other less sophisticated entities might try getting by using redundant T-3s. Don't find out the hard way that your data center provider failed to adequately forecast capacity and can't quickly scale.
2: Redundant power
Many data centers have online electrical backups. UPSes, in other words. If your organization maintains business-critical systems that simply can't go down, be sure that the data center has a second electrical backbone connection. Only N+1 power grid connectivity, to a secondary electrical source, can help protect against catastrophe.
3: Backup Internet
Just as any quality data center will maintain redundant power sources, so too must it maintain secondary and tertiary Internet connectivity. Buried cables get cut. Overhead cables fall when trucks strike poles. Vendors experience network-wide outages. Only by making sure that multiple tier-1 Internet provider circuits feed a facility via fully meshed backbones can IT managers rest assured they've done what they can to eliminate potential downtime.
4: Automatic hardware failover
Redundant power, Internet, and even heating and cooling systems are great, but if they're not configured as hot online spares, downtime can still occur. It's critical that data centers employ redundant online switches, routers, UPSes, and HVAC equipment that automatically fail over when trouble arises.
5: Access control
The importance of physical security can't be understated. Commerce could be significantly affected if just one unstable individual were able to drive a large vehicle into a busy and sensitive data center. That's why it's important that a data center's physical perimeter be properly protected. In addition to physical access controls (keys, scanner cards, biometric devices, etc.), care must be taken to ensure that, should someone gain access to a data center, individually leased sections remain secure (thanks to additional physical access controls, locks, cages, rooms, etc.).
6: 24×7x365 support
Data centers must be staffed and monitored by properly trained technicians and engineers at all times. It's an unfortunate byproduct of today's pressurized business environment but a fact nevertheless. Systems can't fail. Constant monitoring and maintenance is a must. Certainly, many data centers will run leaner shifts during off hours, but telephone support and onsite assistance must be always available. Further, data center services must include customer reporting tools that assist clients in understanding a center's network status.
7: Independent power
Data centers must have redundant electrical grid connections. That's a given. And facilities must also maintain their own independent power supply. Most turn to onsite diesel generators, which need to be periodically tested to ensure that they can fulfill a data center's electrical requirements in case of a natural disaster or episode that disrupts the site's other electrical sources.
8: In-house break/fix service
One of the benefits of delegating services to the cloud is eliminating the need to maintain physical and virtualized servers. OS maintenance, security patching, and hardware support all become the responsibility of the data center. Even if an organization chooses to co-locate its own servers within a data center, the data center should provide in-house staff capable of maintaining software and responding to hardware crises.
9: Written SLAs
Any data center contract should come complete with a specifically worded service level agreement (SLA). The SLA should guarantee specific uptime, service response, bandwidth, and physical access protections, among other elements. Ensure, too, that the SLA or terms of service state what happens if a data center fails to provide uptime as stated, maintenance or service as scheduled, or crisis response within stated timeframes.