This is an article about the application and capabilities of the virtual platform we offer, our first true ‘Cloud’ platform.
The platforms we offer can be generalized into three families:
… and sure, this is an exceptionally broad brush I am using.
To start this story – you need a beginning:
Once upon a time the progression was quite simple. Shared hosting, virtual machine, dedicated server, high availability cluster of servers, geographic balancing. Easy life.
This was a pattern of progression that existed in a world where servers would draw an amp or more each, and you would run out of power in a rack way … WAY before you ran out of space.
These days, the lines are blurred, and with servers being more likely to be drawing a third of an amp for significantly more performance, the progression has changed.
It can pretty much be narrowed down to the following statements:
“Shared – one size fits all – cost effective delivery.“
“You want full control over the configuration – you need dedicated or VM.“
“If you need LARGE amounts of storage – you need dedicated.“
“If you need small, robust and resilient deployments – you need virtual.“
Sure – you can build the dedicated server of your dreams (for a price tag to match and a commitment) – and in which case you can be as fast as you like – but with the current trend for self-contained small deployments – almost containerized or appliances – then suddenly a VM makes all the sense.
The key to our new VM platform is the ability to distribute copies of the file system across multiple locations. While processing will ideally be local to one of the copies, it does not have to be.
The interconnects making this possible are all at 10 Gigabit Ethernet – outstripping the write performance of the collective drives, and further reducing the latency.
Files system and processing can (usually) be hot migrated meaning that you not only have the resilience of data* – but also of processing power.
* In the same way that RAID is not a replacement for a good backup strategy – neither is a remotely replicated drive. We offer a daily snapshot (where possible) for the VM platform allowing complete point in time recovery of the entire instance. However should you require a more granular recovery, or have compliance needs – we would recommend the deployment of an R1 solution.
With current hugely capable processors, and blazing fast RAM – our benchmarks and experiences have shown that these are very, VERY competent performers – when compared against both our earlier VM offerings, and our entry level budget dedicated servers.
In a nutshell: As current trends come full circle back to small, agile, fast deployments for servers (without the world of all the different SCSI cabling!) – If you are thinking of smaller deployments in the 50 to 150G range of storage – unless a very specific architecture is a driver – I would without question suggest our SSD based Cloud platform and a Windows or Linux VM.
Recent articles on this platform:
– VM performance – Phoronix test suite;