Disaster Recovery as a Service and VMware Site Recovery Manager

At VMworld 2011 USA there were quite a few sessions on Disaster Recovery using Site Recovery Manager 5.0. Also lots of focus on Disaster Recovery As a Service (DRaaS). In this posting I will give information on the DRaaS offering by four USA based providers.

DRaaS is basically outsourcing of the Disaster Recovery infrastructure and operations to a Service Provider. Cloud DRaaS  is another name commonly used in marketing for DRaaS.

An organization which wants to protect their IT-infrastructure for disasters (fire, flooding. hurricane, major hardware or software failures) has two options:

  1. build and operate their own failover site;
  2. oursource the failover site to a service provider;

Do it yourself DR is very expensive. Think about having to buy and maintain additional server hardware, storage, networking and software licenses. And having to rent or buy additional datacenter space, cooling and power. And manpower to operate it.

An alternative to DIY/on premises DR is usage of services of a provider/commercial party. Not so many years ago we sent tapes to a recovery site. Once a year organizations performed a test recovery at the recovery site only to find out the recovery failed. Using storage replication made things much easier. However the service provider needed the same brand of storage to be able to replicate the data. This was not very cost effective for the provider as he needed to have all sorts of storage to be able to replicate data from the customers datacenter.

Site Recovery Manager 5.0 changes that. It offers two methods to replicate data:

  1. traditional array based in which the storage array is responsible for the replication.
  2. host based replication or vSphere Replication as it is called by VMware

vSphere Replication is storage agnostic. That means it does not matter which brand of storage array, disks or fabric is used at the protected and recovery site. It can use any storage as long as it is supported by VMware. A component in the ESX kernel takes care of tracking which data has changed and sents this over the network to the DR site. The exact working of vSphere Replication is explained here.  Jason Boche wrote an overview of disadvantages of vShere Replication compared to array based replication in this posting.

The cloud makes disaster recovery much more cheaper and efficient than do it yourself DR. The costs of hardware, software, facilities and management of the DR site is devided over the customers of the service provider offering the DRaaS. It makes sense to share an infrastructure and pay only for the resources you are using (ie Cloud). DRaaS can be seen as a first step of organizations towards cloud computing.

One of the sessions at VMworld about SRM was titled ‘BCO3336: Disaster Recovery to the Cloud: Service Provider Perspective’. This session was more of a sales session than a deep dive technical session. Four US based service providers were on stage and each were given around 10 minutes time to do their sales pitch. All four (FusionStorm, iland.com, hosting.com and VeriStor )   are using SRM 5.0 for their DR to the cloud service offering.

I was surprised to learn that DR to the cloud was not yet offered by many service providers in the USA. All four just started the offering. VeriStor  was presented with VMware’s “First to Market with Disaster Recovery to the Cloud” award during VMworld. The four providers mentioned are all using VMware SRM 5.0. Some others are considering other tooling. Zerto Virtual Replication got a lot of attendance during VMworld.

FusionStorm has two offerings: standard which is a low cost protection per virtual machine. Customers do not have acces to the SRM console. VM’s are running in a multi-tenant infrastructure. vSphere Replication is used. In the advanced offering the customer has dedicated vSphere hosts, has limited access to the SRM console and both vSphere Replication 3rd-party replication is available.

VeriStore reports that the costs for on-premise DR is around $320.000 for 36 months. This includes hardware, software, facilities and operations costs for a SMB infrastructure having 5-10 hosts, 25-75 virtual machines and 5-10 TB storage. 
Cloud DR promises a cost reduction of 40-60% compared to on-premises DR. Costs for standby fee are $4000 and for failover fee $ 7,700 . For 36 months the costs are $ 144,000

hosting.com shows a self service webportal which can be used to test recovery, initiate a failover or failback. It also gives an overview of the status of the virtual machines level of protection. It shows the progress of a recovery and logfiles can be used to find out why a recovery has failed.

 iland.com talked about one of ther customers Psomas who was able to reduce OPEX by 35% by using iland.com Continuity Cloud.

Add a Comment

Your email address will not be published. Required fields are marked *

Current ye@r *