As interest in modular turnkey data center infrastructure grows, deployment challenges remain.
IT managers have more options to expand or build up new data center infrastructure, thanks to the related mega trends of a need for speed and system bundling.
The need for speed is evident in business. With the Internet breaking down traditional distribution barriers, corporations must move fast to keep up with competitors. To meet that goal, an enterprise's computing infrastructure must be flexible.
In response, hardware vendors are moving away from standalone devices, which data center IT teams connect themselves, to integrated turnkey systems that drop into the company network.
Vendors, and teams of vendors, deliver these turnkey data center products to speed up deployment, but executives still grapple with the systems' features and complexity.
Taking IT all in
Converged systems, which combine server, network and storage functions, represent the first generation of turnkey IT devices. The approach has proven to be quite popular: market research firm Technology Business Research Inc. expects spending on converged systems to reach $21 billion by 2019 and be close to the spending on standalone systems.
Phase two is the movement to hyper-converged IT systems. Here, suppliers bundle the underlying hardware as well as virtualized and container software needed to get the turnkey box up and running quickly. Interest in the hyper-converged infrastructure area is also high.
"Customers are adopting hyper-converged systems for a wide range of use cases," said Christian Perry, principal analyst and practice manager at Technology Business Research.
A third phase has also begun to take shape. Here, vendors expand the reach of turnkey data center systems to power and cooling solutions. In fact, the Uptime Institute has seen a major shift in how IT managers view their energy usage in the past decade. Currently, more than four out of five (82%) IT managers put metrics in place to measure IT energy efficiency.
Containing IT services
While the interest in turnkey data center products is high, adoption is low because suppliers have not been able to develop suitable packages. A few years ago, data center equipment vendors, like Cisco, Hewlett Packard Enterprise (then just HP), IBM and Oracle -- along with startups such as AST Modular and Cirrascale -- made a foray in the turnkey space with containerized systems.
Containerized data center vendors take standard 20-foot to 40-foot shipping containers and pack them with servers, switches and other IT and infrastructure equipment, like power and cooling systems. The idea is to quickly add extra capacity to an existing data center or perhaps bring computing capacity to a remote location, such as an oil exploration site. The systems include server racks that each have their own enclosure, so the temperature for each one is set individually, if required.
The goal of data center containers is to reduce the time needed to upgrade an existing facility or build a new one, a process that can exceed 12 months of construction. The new turnkey systems promised to cut deployment time to 90 to 120 days.
To date, the container solutions have garnered niche acceptance primarily for emergency disaster recovery or IT power in remote locations, but not for primary data centers. One reason is that the systems are too cookie-cutter to meet most companies' needs. Data centers typically support a variety of devices, and the container systems are not that flexible, usually supporting one supplier's hardware.
Complexity has been another barrier. IT systems are complicated. Even with the components bundled, a lot of integration work is needed to tie a container system into the current data center infrastructure. Often, the enterprise lacks the needed expertise in-house and must hire third parties to train its staff.
Similar issues occur when troubleshooting these systems. The available tools do not offer the level of visibility that data center techs need to quickly and accurately pinpoint problems.
Bringing IT and data center together
So, what options do enterprises have as they expand existing data centers or build new ones? They can continue to roll their own infrastructure; however, this process is becoming more challenging as demand for IT system resources grows and deployment times diminish.
Web-scale companies such as Apple, Facebook, Google and Microsoft rely heavily on technology to run their business. They have invested billions of dollars in IT systems and crafted varying processes so they can quickly build out massive data centers in a turnkey fashion with generic commodity systems.
But most organizations do not have the expertise or the interest to complete such work and often turn to their hardware suppliers for help. While turnkey IT systems such as hyper-converged infrastructure are simpler to deploy, these vendors are less adept at dealing with power and cooling systems. The power and cooling vendors, such as Johnson Controls and Emerson Network Power, face a similar challenge; their expertise is in energy issues, and they are less proficient when working with compute resources.
Alliances are forming to bridge the chasm. Emerson Network Power has been working with IBM to link Emerson's Trellis data center infrastructure monitoring suite to IBM's IT service management products, bringing Trellis energy management to IBM users. Schneider Electric connected its StruxureWare for Data Centers solution to HPE's HP OneView monitoring and management solution.
Third-party specialists -- CoreSight and Digital Realty Trust, among others -- offer businesses one more alternative with prebuilt data centers nationally in the U.S. and globally. IT organizations can move their applications from on-site to the vendor's turnkey data center facilities. Typically, these suppliers get generic systems up and running quickly, but have trouble supporting system specific requirements.
What can today's DCIM software tell you?
IT teams embrace hyper-converged platform
Steel mill turns into data center with containers