How to Choose the Right IaaS Provider

When shopping around for an Infrastructure-as-a-Service provider, it’s sometimes hard to see the forest for the trees. There’s a lot of talk on GB’s, Ghz, RAM, VM’s and IOPS, and a lot of spin around multi-tenanted, single-tenanted, dedicated resources, uptimes, hypervisors, desktop delivery and storage tiers.

It’s a lot of information to process, and it can be hard work when comparing various providers. It’s difficult to find the middle ground so that you don’t overspending on over-spec’d resources that won’t see capacity, but also don’t underspend on a service that can’t meet your performance expectations, resulting in disgruntled end-users or more unplanned expenditures for upgrades or contract exit.

For the purposes of this article, we’re putting performance aside. Speed of the remote desktops and application delivery should be a given, and most providers offer a trial period to give users a chance to test their hosted environment with a full load before committing to them for the long term. Resources can be tweaked if necessary, and ultimately if the provider can’t deliver on speed and performance, there are larger issues at play, most probably in their overall IaaS design.

So what is the best advice we can give to you when testing the market for IaaS quotes? It’s this:

Define your Recovery Time Objective and Recovery Point Objective for each key business application, and ask each provider to design the hosted infrastructure to meet or exceed these requirements.

Here’s an example of what that looks like from a recent client:

“Our key business requirement is to ensure our ERP Application is never down for more than 30 minutes (RTO), and we never want to lose more than 15 minutes of data (RPO) at any one time.”

Everything else in the hosted infrastructure design is secondary (and sadly, some of it is smoke and mirrors in IT). As long as you can recover within an acceptable timeframe and sacrifice only the work that you’re prepared to re-do you can be confident that your IaaS provider (or managed services provider) will meet their primary obligation.

Look for evidence that the companies quoting you can actually deliver the Recovery Time and Recovery Point objectives that you need to keep your company operating smoothly, even if you are under attack from malware. And if you choose to keep servers on-premises, be realistic about how you’re going to ensure the same recovery objectives through a Managed Service provider.

So what do you look for that will assure you of their ability to deliver on their primary purpose?

1. Do they include Disaster Recovery as part of the service, and how often is it tested?

At its core, providing infrastructure is just a matter of specifying, procuring and maintaining hardware (or virtual appliances) to store your data. It’s not until that data is placed at risk that the true cost or benefit of your choice of provider is revealed. If your provider offers the first part, but not the second, what value are they really bringing to the table? (And how much does the disaster recovery component cost?) You need both services in an IaaS proposal to have a complete picture.

However, even the best DR plan needs to be regularly tested to ensure that it will work when required. This is another reason why your potential IaaS partner should include DR as part of the Infrastructure proposal.

2. Where do they store your data and across how many physical locations?

There are more than one hundred data centre providers in Australia ranging from small to enterprise-level. Quality of Service varies markedly between them. Your providers should be able to name each location that will handle your data and explain why it is the best choice for your company’s Value Point (the point where your needs and your budget cross over).

As part of a solid IaaS design, there should be a Primary data centre, backed up to a physically separate Secondary data centre, as well as a regular offline copy of the data performed to tape or disk for archiving purposes and stored in a third location.

Pro-tip: when it comes to data centres, bigger is usually better. If your primary or secondary infrastructure location isn’t a name brand, your data could be at more risk than it should be.

3. Will the IaaS Provider define the recovery times and points for each application and server, and will these times form part of their Service Level Agreement?

The design of the IaaS and the method by which applications and data are backed-up and replicated are critical to the recovery times. If you request no more than 15 minutes of downtime for any application, this will be more expensive to maintain than a platform that delivers, say, a maximum of 1 hour of downtime. This is why recovery times and uptimes are important to note when comparing ‘apples-to-apples’ amongst IaaS providers.

The service provider should guarantee to meet their proposed SLA’s, as well as regularly test recovery times as part of Disaster Recovery planning.

4. Have they designed-in multiple pathways to get your data in and out?

Once your infrastructure is hosted offsite, suddenly your data service becomes critical.  This makes it one of the weakest links in your Business Continuity chain. When a contractor cuts through a cable in a nearby street, what secondary measures come into play to handle your network traffic while remediation occurs?

A good Infrastructure Provider should ensure that you can’t be cut-off from your data through a single physical act, by having multiple carriers and services to the data centres, as well as your offices. This can be achieved through a number of cost-effective methods and should be included in the main contract price, not as an add-on.

5. Ask: Who bears the risk for failure to deliver?

If the infrastructure fails, and you lose a day’s worth of productivity across your organisation, who bears the risk of cancelling the contract? If you need to give between one and three months of notice, there’s a big disincentive to cancel. How is your IaaS partner putting their money on the line to guarantee they will deliver?

Concentrate on the RTO and RPO and you’ll find it much easier to make a decision on your IaaS provider.

But is IaaS itself right for your business?

Transitioning to IaaS might not actually be the best move for your business!  You don’t have to move entirely into a hosted infrastructure arrangement in order to benefit from the cloud. If the cost of transitioning to a full cloud environment is roughly the same as maintaining on-premises servers with an off-premises backup and increasing the capacity of your WAN link(s), that’s still a valid solution moving forward. Companies with less than 30 staff should consider this option alongside full cloud solutions.