We are pleased to be able to announce the launch of the new MD Web Hosting site. Our staff have worked hard over several months to design, build and test a unique website that is easy to navigate and shows off our great new line of products. We would like to hear feedback about the new design or anything else in this blog.
Why are we starting a company blog after all this time? Well, our client base has grown leaps and bounds since we started hosting. MD registered its 50,000th domain not too long ago, and we are well on track for the 60,000th. Blogging is a great way of letting all our customers know what is going on. We are also using our Blog to introduce the secret product that has been in the pipeline all year - codename project Cluster.
Clustered Hosting
As many of you know, there are many problems with the standard shared hosting model. Servers break, hard drives fail, and tears flow! Being in the front line, we know very well the difficulties of maintaining a large server fleet, and the hassles caused when hardware failures bring customer sites down. There are several points of failure in the single server “silo” model.
We decided to fix this once and for all. After many months of brainstorming, planning and serious capital investment, we have released our new Clustered hosting product.
How does clustered hosting work?
The principal aim of clustered hosting is to remove any single points of failure. Rather than running a website on one lonely web server, we have spread the workload among a cluster of servers. Sites are dynamically transferred between web-servers in real time and with no downtime. The cluster knows when a server is under high stress, and will distribute excess load to other members of the cluster for smoother performance. Compare this to certain budget hosts that cram their web servers to the brim - load balancing those disasters can take weeks and cause a lot of heartache for clients.
Should a server fail for any reason (and believe us, all hardware fails eventually) then our resource scheduler will rescue all the websites from that server and restart them on different servers with spare capacity.
Storage Arrays
At the heart of our clustering infrastructure are the enterprise level storage arrays. Rather than risk storing data on single servers or drives, we have created logical drives on massive arrays of high performance SAS hard drives (more commonly known as Storage Area Networks - SANs). A SAN is highly fault resistant and self healing - it can suffer multiple hard drive failures and rebuild these from hot spares in minutes. In fact, every active component has a spare to take over automatically in case of failure. A SAN is the ultimate storage device, but because of the large capital outlay required they are typically only seen in banks, financial institutions and other blue-chip enterprises.
Don’t be wooed by promises of redundant “RAID1″ either. RAID does improve availability in its own limited way, but it is certainly not a reliable data protection strategy in isolation. Servers are sometimes taken offline for days to rebuild a RAID set after drive failure - and that is assuming the paired disk has no faults or bad sectors.
Backups
We are also excited to be able to offer backups for the first time. Using the latest in data storage techniques and hardware, we can store a full daily backup of all your data. While that may not sound that exciting, this next part is - after the first daily, we can continue making full daily backups with very additional little resource usage, which allows us to store a massive amount of data. We can also browse back in time, like a time machine, to recover the data exactly as it looked like on a particular day. As far as we are aware, no other web hosting company in Australia offers this level of data protection.
Summing things up
The technology discussed is only one part of our “Unified Business Continuity Strategy” (UBiCS for short!). In coming posts we will discuss other crucial areas of UBiCS and how it helps your small business or website.
FAQ - How do I sign up to Highly Available hosting?
We should mention that the clustered, high availability hosting and backups are only available for our new business class packages (Business, Premier and Ecommerce). All new Business customers will be provisioned on the new cluster. Existing customers will have to request an upgrade to the next applicable plan, but are welcome to stay on their existing plan if they want. US customers will have to transfer their website to Australia to be eligible.
Upgrade path to the new hosting cluster
Budget, Bargain, Starter, Dingo => Business
Kangaroo => Premier
eShop => Ecommerce
RSAU5 => Developer
Filed under: Company Blog posts by admin
No Comments »