Attended a VMware presentation yesterday, several items which I was already aware of were discussed on more detail. A lot of these lead to some interesting options for the University, in no particular order:
- Lab Manager – http://www.vmware.com/products/labmanager/ Cool tool for Computer Science/Software Engineering
- Stage Manager – http://www.vmware.com/products/sm/ possible good for LIS/ACU to manage introduction of new services
- Continuous Availablity – coming soonhttp://www.vmworld.com/vmworld/message/1051 Real Time B.C. solution
- Storage Mirror – another “future” – mirrors Virtual Disks across LUNs in real time – fits somewhere between EVA storage mirroring (HP Continous Access) and application level storage mirroring (SQL Server 2005/Exchange 2007)
- LUN Balancing – using Storage VMotion in a similar way to Vmotion is used by DRS at the moment to load balance storage I/O loads across LUNs, and therefore across Disk Arrays
- VDI – http://www.vmware.com/products/vdi/ Virtual Desktops – ideal to deliver “fat clients” and applications via a web browser/thin client – think virtual Open Access Machines available via a browser in a Hall of Residence or over the vacations from home or a LIS/ACU Managed Desktop accessible to Solaris/Mac/Linux users.
- DPM – http://www.vmware.com/products/vi/vc/drs.html Using VMotion to reduce the number of ESX servers in use when the system is lightly loaded (e.g. at night) and then powering off surplus ESX servers.
- ThinApp – competitor to Softgrid
What’s begining to come clear is that VMware is a very powerful system from several strategic view points:
- It’s Green – or put it another way, LIS’s machine room for one has power supply and air conditioning issues – VMware helps reduce our power requirements – clearly if this is applied University wide then more power savings are easily possible.
- Load Balancing (CPU, RAM and now Storage) maximises performance
- Continuous Availability, HA, DRS, SRM are very powerful tools which provide serious (and easily obtained) solutions to both Business Continuity and Disaster Recovery
In order to maximise the Unversity’s investment in VMware, it clear that the more it’s clear that a single “Cloud” managed by a single “Infrastructure Team” (analogous to the single Network and Telephony Teams) would give the University a far more flexible and powerful system than the current distributed “Silo” approach.
So what would this “Cloud” look like for the University? Here’s a couple of ideas…
- It would spread over 2 or more machine rooms on the Singleton site to with mirroring providing continutity of service for critical systems.
- It could use a 3rd party Vmware hosting solution or the second campus to provide a full site disaster recovery solution.
- The SWWEP members (SMU and TCC) could copy our solution, using perhaps our second site or the same 3rd party hosting provider for full D.R. They could even consider SU managing and providing their “Cloud”?
- Anyone within the University would have access to this “Cloud” to provide them with hosted servers or PCs for research or departmental projects – this would probably save the University substantial costs:
- It would reduce the number of “Servers under desks” and second PCs for Research/Testing (which use Electricity even when the majority are low use systems)
- It would reduce the number of machine rooms needed – saving on air conditioning costs.
- These servers would benefit from serious business continuity and distaster project procedures, safe guarding important and valuble research materials.
- All the major service providers (LIS, ACU etc) as well as departments with heavy investment in I.T. (Computer Science, Engineering, Media Studies, Health Science, Medical School) would all “buy in” to this central service, saving the Unversity a small fortune (in server hardware, electricity, software licencing) and allowing the I.T. staff in these departments to focus on supporting their specific use of I.T.
- A Unified Comms Systems (Telephony, Email, Instant Messaging, etc), as it heads towards being VOIP based, could benefit from building on this “Cloud” being protected against loss of a machine room, or even the whole campus.
Of course I could continue on this line of rationisation by pondering over why the University tolerates multiple E-mail Solutions, multiple Active Directory installs with several groups duplicating efforts to provide managed desktops etc etc. This sort of local “control” of systems may seem important to those in charge, but to end users – staff and students – they look very poor – why can’t I share my Calendar with you? why can’t I log into Engineering with the same account details as I use in the Library? Why do I get passed from pillar to post when I want help to login to service X? etc etc
Obviously it takes a lot of political will to get people to “let go” of their own solutions, but surely it’s better for a School or Department to look at its own I.T. staff and say, what jobs can I off load to a central service, so they can help develop and support School based solutions for School based problems?