Data center build-out considerations for small IT consultancies
By Erik Eckel
April 19, 2013, 8:29 AM PDT
Takeaway: Your IT consulting firm can gain a competitive edge in its region by building its cloud network. Here are three factors to consider before making this investment.
Basic technology services — server monitoring and administration, desktop support, network troubleshooting, and end user help desk services — are sufficient to power small consultancies. Consulting businesses wishing to build competitive operations that scale as customer needs evolve, however, must consider building their own cloud infrastructures.
Firms partnering exclusively with the GoDaddys, Carbonites, Mozys, and Rackspaces of the world will find service offerings and revenues limited. Only by investing in their own cloud services will consultancies prove capable of building a strong regional presence. Here are several considerations smaller consulting organizations seeking to grow by developing their own cloud networks must take into account when making such investments.
Data center selection
A consultancy’s cloud services are only going to prove as robust and reliable as the data center in which those services are hosted. When selecting a data center, consultancies should seek the most dependable partner available. Raised floors, conditioned electricity, subfloor power distribution and telecommunications connectivity, redundant electrical systems, biometric security, 24/7/365 monitoring, and climate and humidity controls are prerequisites, minus any element a data center will prove insufficient for meeting the consultancy’s cloud needs.
When selecting a data center, consultancies should confirm the infrastructure provider they choose is reasonably protected against unplanned outages and network interruptions. Before a consultancy can put email hosting, web hosting, antispam, and remote backup systems in place with confidence for clients, they should confirm the data center offers redundant backup power, redundant Internet connectivity, zoned fire suppression systems, multiple power grid supply, and sufficient bandwidth via multiple OC-X connections.
Rack inventory
Significant time is typically spent selecting and researching a data center. Additional energy is usually invested negotiating service level agreements (SLAs) for the electrical and Internet connectivity services the infrastructure provider will deliver. Resources dedicated to forecasting required rack space are often shortchanged.
If a data center enables the consultant to simply add half-rack allocations as circumstances require, that’s awesome. Administration usually proves most efficient, though, when server racks remain contiguous. No consultant will appreciate having a half dozen half-rack setups distributed throughout a wide-ranging data center, but that could happen. If a consultant begins by securing a half rack, the remaining space in that rack may be sold to another firm by the infrastructure provider, while other vacant adjacent rack space may be sold to another data center client. Thus, consultants will be well served better forecasting not only how much rack inventory their cloud services will require but also any co-location and other data center services they might need.
In addition to calculating space for servers, consultants must remember to reserve space for telecommunications equipment, routers, security appliances, and storage area network (SAN) hardware. A standard server rack is 42U, with a single U typically translating to 1.75″ in equipment height. A single SAN or robust server can occupy as much as 4U, or almost a quarter of a half rack, on its own.
Bandwidth allocation
Estimating ideal bandwidth is tough. There is no formula consultants can employ to perfectly determine peak usage. As cloud services expand, consultants should expect to host websites, email services, antispam equipment, and even image backup and virtual machine snapshot storage in the data center. Trying to calculate how much bandwidth is required to enable proper data flow in and out on any given day is difficult.
Instead of attempting to calculate a single, static number, such as 15Mbps, 20Mbps, or 50Mbps, consultants should leverage the assistance and flexibility of the data center’s engineering staff. Most infrastructure providers now offer dynamic pricing. As long as the consultant words client co-location contracts to support dynamic bandwidth pricing, the consultant’s data center services should prove capable of dynamically supporting a client who experiences web or other traffic spikes, in which case the data center’s burstable bandwidth services can be leveraged and the resulting bandwidth spike can be billed to the client at predetermined contracted prices.
The same strategy holds true for the consultants’ own cloud services. As the consultants’ service offerings and supported client base grow, so will the need for more bandwidth. Consultants should partner with a data center provider that can adjust bandwidth allocations as needs change.
Engineer access
Cloud- and co-located services place heightened demand on preventive maintenance. When problems arise with cloud-based and co-located services, repairs must be delivered immediately, as by their nature larger volumes of users depend upon the data center-hosted services.
For these reasons, consultancies must plan in advance which engineers will receive access to the data center. Further, consultancies should develop plans to ensure an engineer requiring access to service cloud-based equipment does not automatically receive access to a client’s co-located server racks. Segmenting access helps maintain proper security controls.
Yet, no consultancy wishes to find itself with a server down and an engineer onsite but the engineer lacking proper credentials or access. Consultants should ensure that a primary and a secondary engineer always possesses proper credentials and access, as the process of obtaining clearance can sometimes require two or three hours.
