"The custom applications developed by PeaceWorks for our company have become the very core of our daily operations, providing an order of magnitude improvement in our overall effectiveness and efficiency... with best in class custom software solutions and a shared perspective of community, ours has been a uniquely satisfying business relationship that we look forward to continuing for many more years to come."
When Backups are Not Enough: Failover Server Design
You're a systems administrator and things are good. You've got backups in place to protect your precious data. Congratulations, this is great! But is it enough?
When we think backups, we generally think of disaster recovery situations. However, what is often overlooked is how much downtime is involved during the backup recovery process. It's great that we can get our data recovered, but what caused our data to dissapear in the first place? And how long will it take to get everything operational again? If your employees can't work, how much revenue will be lost? This is where looking at Failover Server Design (also known as High Availability, or HA) comes into play.
In a typical setup, after failure, a new server needs to be built, Operating System & applications need to be installed, and, of course, data needs to be restored from backups. This can be be costly and time consuming. Often times people who rely on the services from that server are left unable to perform their duites. This becomes a downward spiral and the costs start growing.
How does HA solve this?
HA solves this by having a cluster of nodes (servers) that each have the ability to run one or all of the services required. In the event of a failure, another one of the nodes can take over the duities and continue to serve with little disruption. This does not require reconfiguration of the client systems. The client continues to operate, often not realizing that there was an issue.
How can the new node get to my data so quickly?
In the HA setup, when data is written to the main node it is then replicated to the other nodes in realtime. This is done using Distributed Replicated Block Device or DRBD. This can be thought of as RAID-1 over TCP/IP. This allows the all nodes immediate access to your up-to-date data. DRBD even plays nice with local RAID-1 setups. So each server can have its own local (hardware or software based) RAID-1 in addition to the RAID-1 via DRBD.
When disaster strikes your infrastructure, you want to ensure the services that your employees rely on are continuously available, including your data. The realm of HA is vast and PeaceWorks can provide consultation to see what levels of HA are right for your company or organization.
- About Us
- Integrated B2B Solutions
- IT Infrastructure
- H/W and S/W Sales
- Repair Centre (Toshiba)
- Online Backup
- News & Events
- Contact Us