We are very excited to announce that we've partnered with a data center in Dallas, TX to host our brand new, fully owned hosting infrastructure. While The Planet and SoftLayer have both treated us very well over the past 8 years, it was time for us to move on. This has given us the opportunity to create an environment that specifically addresses our growing needs. We will be migrating all clients to new servers over the next few months. Notifications concerning these migrations will be posted here in the announcements forum and will be sent out via notification tickets in our portal.

While the hardware upgrades are a very significant part of this, other features are also being introduced on all new servers to further improve your overall hosting experience. For our hosting resellers, this will give you even more ways to differentiate yourself from other providers. A few of the more important changes are described below.

Hardware Upgrades
Admittedly, this has been one of the most exciting parts of this whole process and we can't wait to get all of our clients migrated to these new servers. Just to offer a point of reference, a typical server of ours currently has a configuration similar to the one listed below. Keep in mind that this build is now just a little over 4 years old and these servers still have no problems meeting demand.

Dual Xeon 5148 (Dual Core, 2.33Ghz) - 4 physical cores total
4GB ECC Registered DDR2 RAM
Adaptec Hardware Raid w/ 128MB cache
2x300GB SA-SCSI - RAID 1

Now lets take a look at our new server configuration:

Dual Xeon E5645 (Hex Core, 2.4Ghz) - 12 physical cores total, 24 logical cores total with HT
48GB ECC Registered DDR3 RAM (easily upgradable to 96GB)
Adaptec Hardware Raid w/ 512MB cache
2x 300GB Intel 320 Solid State Disks - RAID 1
12x 1TB Western Digital RE4 SATA Disks - RAID 10
2x 1TB Western Digital RE4 SATA Disks - Hot Spares

It is easy to see the huge increase in processing power and system memory (RAM). You'll also notice that we've stopped using SA-SCSI disks entirely. These actually worked very well for us but times have changed and we decided on an alternative solution. Instead, you'll see we now have a couple of solid state disks in there. These will help to eliminate disk performance issues in key areas (such as MySQL). The pool of SATA disks in RAID 10 will provide a much larger storage area but more importantly at 12 disks the array is quite fast. And finally we have 2 disks configured as hot spares so that if a disk does fail the RAID controller can automatically and immediately rebuild the array on another drive.

All of our new servers will be running CloudLinux. We could write about this for many pages but we'll try and just hit some of the key points here for now and save the rest for another forum post.

In a typical shared hosting environment there are no hard resource limits for each user. Because of this, it is easy for a single domain to use almost all of the available CPU and RAM on the server. We monitor all of our servers very closely to make sure this doesn't happen, but ultimately it is still a monitoring solution such that correction has to be taken after the incident has already started.

This is where CloudLinux comes in. CloudLinux creates a lightweight virtual environment (LVE) for each user. Each user can then be limited to a specific amount of CPU and RAM. When a user reaches their limits, their requests may be served a bit slower, or in more severe cases the server may stop serving their requests entirely until the cause of the high usage subsides. This comes in very handy in cases of out of control scripts or bots temporarily flooding a website. In either of these cases, CloudLinux would limit the impact to just that one user versus having the incident possibly impact the entire server.

Best of all, CloudLinux records this usage and makes it visible to all users directly via their cPanel. You can even create and view graphs of your usage over a specified period of time. This helps take a lot of the guess work out of system resource usage for shared hosting accounts.

Ksplice Uptrack
Contrary to popular belief, Linux servers need to be rebooted for much the same reason as Windows or OS X computers do for updates. Previously, whenever a kernel update was released, we had to install the new kernel and reboot every server in order for the update to take affect. Ksplice plays one very important role in this: it allows us to apply the kernel updates without rebooting. This means that we can apply the updates quickly and then if necessary schedule a reboot for a time that is less likely to impact users.

RVSiteBuilder Pro
We've had some requests for this over the years and finally all of our new servers will include RVSiteBuilder Pro. This will be accessible directly via each user's cPanel and will allow users to easily create their own website. With over 900 templates available, there are a lot of different options to choose from. For those that already offer web design services and don't want their clients seeing this, you can simply disable it via the WHM feature manager.