Colo vs. DRaaS: Which DR Solution is Best for Your Company?

David Davis is the author of the best-selling VMware vSphere video training library from TrainSignal. He has written hundreds of virtualization articles on the Web, is a vExpert, VCP, VCAP-DCA, and CCIE #9369 with more than 18 years of enterprise IT experience. His personal blog is www.VMwareVideos.com.

In my career as an IT Pro, I worked as a network, server, and virtualization admin then later as an IT Manager. I worked for a privately-held family-owned company where we constantly felt pressure to provide low-cost solutions that offered high value for the company.

I've always been a "do it yourself'er" and never felt comfortable putting too much faith in service providers, if I didn't have to. Keeping with that mantra, when it came to a disaster recovery solution my company (with my recommendation), opted for a self-managed co-location option instead of using a DR option from a service provider. Why? The primary reason was to maintain control and because I felt that we could keep the cost down.

However, since that time (about 5 years ago), cloud-based disaster as a recovery (DRaaS) solutions have taken shape and become viable options for disaster recovery. Many people may think that DRaaS is only for the largest enterprises or companies with massive Internet pipes but neither of those statements are true.

Capex vs. Opex

Before I breakdown the resources and costs related to implementing a colocation facility to DR, first understand that disaster recovery is a form of insurance. No one is used to buying insurance in one large payment. If your insurance agent said that they could sell you an insurance policy valued at $1 million for just $500,000 upfront, you would laugh as them.

Instead, the acceptable and less painful way to buy insurance (or DR services in this case) is a flat monthly rate tied to your consumption, or level, of protection. Using a pooling and economies of scale approach, insurance services are similar, in pricing model, to disaster recovery as a service (DRaaS) services because of their “pay as you go pricing”.

I’ll guarantee you that your company’s financial officer (CFO/controller) would be much more likely to approve any disaster recovery plan that includes simple, pay as you go pricing, tied to usage instead of a colocation-based plan simply because of the “operation expense” (opex) vs. “capital expense” (capex) model.

Colo vs. DRaaS

When I set about to help my company create a colo for our disaster recovery site, we did everything we could to keep the costs low but the basic pieces, and their costs, were:

  • Servers – we needed either a 1:1 mirror of all our servers or at least a 1:1 mirror of our most critical production servers. This was made easier thanks to consolidation with virtualization and the hardware independence that it provided. However, there is still at least a cost of $2000 per server, minimum, capital expense.
  • Storage – perhaps one of the most expensive capex items, you’ll need either an identical SAN used in your production datacenter or a subset of it. If you want to do synchronous data replication, that is likely going to require an expensive software license. Storage may cost from $10K to $100K+.
  • Network – adequate Gig E network switch to connect all hosts, VPN/firewall/router for Internet/WAN connectivity. Estimated cost $10K+ depending on how many servers you have.
  • VPN or dedicated circuit – you need a way to connect your primary datacenter to your backup site. You could use a secure Internet VPN or a dedicated WAN circuit (like MPLS) at primary and secondary datacenters. Depending on the site of your bandwidth and the distance between the datacenters, estimated cost is $2000+ per month.
  • Backup Data – you will need your backup data at the DR site to restore when a disaster occurs. This could arrive there thanks to data replication or offsite backup storage (with costs varying widely based on your solution).
  • Colo facilities (datacenter floor space, power, Internet bandwidth, cooling, rack, and cage) – one of the most expensive pieces of your colo DR solution is the facility you select and the cost will vary greatly based on the number of servers you’ll need to put in the colo.

Potential issues with using colo include:

  • Can’t start small - you can’t just protect 1 or 2 servers, the colo facility will be built to protect a minimum number of servers
  • Scalability – you can’t add 20 more servers overnight, for example, because you have to order them, take them to the datacenter, rack them, and configure them.
  • Initial capex – the amount of equipment that must be purchased up front is too large to easily get started and it too painful for many companies to absorb. Because of that, too many companies simply don’t create a useful DR site.

With a DRaaS solution, you’ll pay based on the number of virtual machines or physical servers that are protected. It’s that simple.

In summary, while you may initially lean to using a colo disaster recovery service, like I did, I believe if you take the time to analyze today’s disaster recovery as a service offerings (DRaaS), I think you’ll find that they are better options as they will offer your company a predictable monthly IT expense (and operational expense) that makes sense to everyone in your company – even the CFO.

More about: