Monday, 29 December 2008
Out of a job?
The year ahead
Saturday, 27 December 2008
Cloud...the real cost
I stumbled across this great post on simple economics of utilising for a period of 24x7x365 a conventional hosted server environment against the cost of utilising cloud compute services i.e. Amazon EC2. This is something I was going to look at doing myself but I outsourced it for Free as Chris Fleck at Citrix has done it for me so I will just add half intelligent comments http://community.citrix.com/blogs/citrite/chrisfl/2008/09/27/Cloud+Economics+101+-+Part+1.
Cloud compute has a great selling point at the moment for lower end SMB's and startups who cannot afford to implement server and backup infrastructure due to a lack of capital fund or investment money. In 2008 I attended a Cloudcamp event http://www.cloudcamp.com/ which was full of buzzing web development and software startups all utilising the Amazon and other cloud offerings to get there show on the road.
When looking into the key technology enabler for Cloud computing with players who were sponsoring the event, for cloud to be so turnkey and cost effective the simple reason is Server virtualisation. Virtualisation is allowing the the Cloud providers to partition and provide what could be high VM to Server ratios (i guess a minimum of 10:1) to host multiple server instances which are predominantly Linux based at an exceptionally cheaper cost base than implementing physical single instance boxes for each customer, being able to utilise a virtualisation layer also includes the per unit chargeback facility which are done by CPU cycle usage per hour of each server instance.
Within the Citrix comparative report they haven't mentioned the use of Virtualisation in both the On premise and Co-lo solution to slash further cost. The cloud already utilises Virtualisation, it has to to keep the cost model down so It does baffle me as Citrix are a contributor to the virtualisation market and they are not even piping there own product within such a report!
I'm not bashing citrix, this is a good report (bar the sales pitches) but what really needs to be looked at before people start running into building an Infrastructure within the cloud world is to ensure that the future growth and planning of that hosted environment is compared and planned effectively against the future business need. If the small start ups and SMB's start becoming heavily reliant upon the cloud due to the short term apparent cost savings noticed when they started up they need to ensure that they either have a solid contingency for investment when the tipping point of the businesses requirements are in effect more cheaper being done by purchasing Infrastructure on capital or lease via an alternative solution i.e. hosting on premise.
It will be quite interesting to see how the Datacenter industry and the Virtualisation players certainly start to market there core products against Cloud computing alternatives, up until 2008 they were very much in war against themselves and the physical server world (which to be fair was easy).
Saturday, 20 December 2008
Oh my god....
Cloud.....OK the last time in 2008 hopefully?
Sunday, 7 December 2008
VMworld 2008 material
The direction VMware is going now seems to be a focus and drive on the core vendor relationships and building bridges with the ISV's like it is going out of fashion, for an Architect this can mean great things to come and better adoption rates for anyone embarking on new virtualisation projects.
Some of the ESX performance related roadmap looks great for taking them up to the next level and above competitors, Chad Sakac has a great keynote session which details and demos at http://virtualgeek.typepad.com/virtual_geek/2008/09/so-what-does-vs.html on the new vStorage API's which utilise on first release in ESX v4.0 EMC Powerpath, ESX definitely needs to move towards changing the current multipathing offering. Another key future feature is the usage of "storage dedupe" which hopefully will provide greater benefits when performing high IO disk operations like Storage VMotion and maybe even better VCB interaction.
VMware FT is something that doesn't look as simple to use and architect as Vmotion/HA, when viewing prerequisites it requires massive amounts of planning for additional supportive CPU and RAM allocation due to the requirement to naturally have a standby VM on another ESX host.
FT will probably become more dominant on adoption one or two releases from ESX 4, it will require a massive change in landscape requirements for any current Virtualisation estates, I will be embarking on a project that will certainly include planning consideration for this future release, however if you need to justify and purchase new ESX hosts, licenses etc for your estate this could be a hurdle.
The key benefits and cost saving with VM FT from less downtime will however pay for this in itself no doubt, certainly when you look at costs of something like Neverfail or there competitor Everrun.
Will follow up on more reviewed content as and when
Subscribe to Posts [Atom]