One of the advantages of implementing a Private Cloud in your organization, is the possibility to build a solid Disaster Recovery strategy using Public Cloud infrastructure. Read on, to find out more…
What are Private and Public Clouds?
A Public Cloud service provider creates common scalable infrastructure (Servers, networking, bandwidth, etc) in their data centers which can be utilized by many individual companies to host their applications. The main advantage of a Public Cloud is the low capital expenditure for companies hosting their data (no need to buy hardware), Pay-as-you-use flexible billing model and the ability to scale infrastructure requirements up or down at any given point of time. Private clouds do pretty much the same thing but the infrastructure is owned by the companies who create the private cloud within their organizations. Of course, server virtualization is an important component of Cloud Computing.
Why is Disaster Recovery required?
When you have a Data Center (DC) which hosts some mission critical/ business centric applications, you might as well want that data center to be working without interruptions as the productivity loss and financial loss is generally quite high even if the data center is down for a few minutes. And data centers can be down for a number of reasons like floods, fire, network (link) failure, power failure, configuration errors, etc. So, a DR (Disaster Recovery) strategy is implemented by companies so that even if their primary data center is down, their applications can still be running in an alternate DC/DR site.
What companies have been doing to create an effective Disaster Recovery strategy?
- Back up their data and applications using Tape Drive/ Disks and store it in an off-site location. If disaster strikes, the data and applications can be reloaded into similar systems in a different location and users can be directed there.
- Replicate their data either in real time or near real time continuously over the WAN network to a Data Center dedicated to Disaster Recovery. The DC for DR is owned by the company and techniques like Bare Metal Recovery might be used.
- Renting servers/ infrastructure in a Co-location facility, run by a Co-location service provider and paying monthly rent for the same while duplicating the data/infrastructure over there in order to quickly switch over when a disaster strikes.
These Disaster Recovery strategies are fine but depending on the method, there are some drawbacks to the same – Some of them are very time consuming, some of them are very expensive (as hardware and software needs to be replicated), some of them require similar server configuration, some of them require OS and patches to be installed and then recovery process started, some of them are prone to manual errors, almost all the three result in inefficient utilization of servers, networking equipment, WAN Links/ Bandwidth, Electricity, Infrastructure costs, etc in the DR site.
How can Cloud Computing enable a better Disaster Recovery strategy?
This involves both Private Cloud & Public Cloud.
First, companies need to get all their applications in their companies to become a part of a private cloud within their company. This has many advantages and one of them is server virtualization. Virtualization basically decouples the hardware from the OS and application so that they can run on any hardware and any number of OS/Virtualization pairs can run on a single hardware device. These are called Virtual Machines (VM) and they allow efficient utilization of server resources including the option to dynamically move between various physical servers, as and when required.
Ok, so now we are already utilizing the computing infrastructure efficiently! Some Private Clouds can burst into public clouds when their capacity has been exhausted. Well, even though they are ‘CLOUDS’, their capacity is constrained to the amount of physical infrastructure already provisioned. With Private Clouds, it is easy to provision additional infrastructure, but it still needs to be done!
The advantage with a private cloud is, the entire data center (if built on a single cloud provisioning mechanism) can be exported as such to another site (Public Cloud) over the WAN Links and these two sites can be synchronized with each other in real time or in a frequency that is acceptable. Note here that we are exporting VM’s (Virtual Machines) which are hardware independent and can run on any hardware, including a Public Cloud.
When disaster strikes, the primary site is totally down but the Public Cloud where the data / applications have been synchronizing (through VM’s) is available and all the users can be manually / automatically directed over there.
What are the advantages of using a Public Cloud for Disaster Recovery?
There is a huge cost reduction. That’s mainly possible because of the pay-as-you-use model of the Public Cloud. A Disaster does not strike very often, so if a Data Center is created specifically for DR, there are chances that it might not be used at all or it might be used for a very short period. That’s a lot of resources getting wasted, but still business criticality might require them.
A company can expect to spend very less for storage/ computing resources in a Public Cloud during normal periods because only limited resources are rented to host the VM’s that are going to lie dormant. And bandwidth transfer is incurred only for synchronization between the Public Cloud and the Primary Data Center. If the synchronization frequency is less (instead of real time), there can be further reduction in the resources consumed/ cost.
When disaster strikes, these VM’s can be expanded in to as many servers as required quickly and can utilize any amount of processor resources/ bandwidth as the Public Cloud enables On-demand utilization of such resources. Now, the company incurs a considerable cost for running its whole data center out of the Public Cloud, but it is still going to be for a short duration. Once the primary data center is up and running (back), the load can be transferred back over there. The infrastructure created by Public Cloud providers are shared with many companies and hence they can be utilized efficiently.
What are the limitations of using the Private Cloud for Disaster Recovery?
Well, till now a very rosy picture was painted but reality is always stranger than propaganda! 🙂
First of all, private clouds are a relatively new concept in enterprise data centers. There are no standards and presently either a single vendor provides limited resources to enable a private cloud or there are too many pieces of the solution that needs to be woven together. Even if a company manages to build a private cloud, the synchronization between private cloud and the public cloud may not (yet) be supported. Or it might be supported by a few vendors in a limited way.
The larger distances between primary data centers and Public Cloud data centers introduce latency for users during fail-over. Public Clouds are viewed as insecure and a lot of companies may not want their data to be exposed to external service providers. A private cloud can be created at a DR site, but its the same as having another DR Data Center. The storage technologies used by the company and the Public Cloud vendor maybe incompatible. The network settings / DNS propagation during fail-over needs to be taken care.
Re-synchronization of data / applications back to the primary site once it is back up and running is another issue. In spite of these reservations, we can be sure of one thing – Private and Public Clouds are going to be the inevitable future of IT infrastructure and if not now, companies will slowly start adopting them and Disaster Recovery could be one of the benefits they can hope to achieve. So, Disaster Recovery should become feasible in the near future using cloud infrastructure, if not already!
You could stay up to date on the various computer networking/ related IT technologies by subscribing to this blog with your email address in the sidebar box that says, ‘Get email updates when new articles are published’