I want to move your attention towards Virtualization for some minutes specially who are directly involved with designing and budgeting of IT infrastructures. It's great if you have already idea and knowledge on it, if you don't I shall request to kill some of your minutes please.
What is Virtualization?
The core concept of Virtualization is adapting multiple guest operating systems on a single hardware platform. These operating systems (either can be Windows or Unix or Mac, everything is supported now) will share the same hardware resources like CPU and Memory as you have allocated for each and they will use upto their limits. It's the core concept of virtualization, but it's technologies are growing day by day, now a days you can have redundancy of guests, moving one guest from one platform to another within seconds over Gigabit networks, virtual database clustering etc. etc. just what all we need at productions. Even the Bigs like Oracle, Sun, Microsoft, Cisco, Juniper all have already adapted their products for Virtualizations streamlines and also released their compatible products. Even processor manufacturers like Intel and AMD already released processors supporting Virtualizations.
So, why Virtualization?
Want to give you a simple example, it might be exactly like your production servers already running. Just point to your network monitoring system (NMS) and look for arbitrary four servers (like HP Proliant DL320) each with 1U rack space. What's their CPU and Memory utilization? I guess the CPU is 10% during cold hours and fluctuating upto 50% at peak hours and the Memory is 30% at cold hours and 90% at peak hour. What about their storage utilization? Ok, let me guess they are using network storage systems. Now, if you tell me to design Virtualization systems at your data center I shall consider the parameters
1. Your rack U space (1U X4 = 4U in total)
2. Cooling power and operations (aKW X4 = 4aKW, a is considered as utilization per server)
3. Server power (bKW X4 = 4bKW, b is considered as utilization per server)
4. CPU utilization (10% to 50% fluctuating)
5. Memory utilization (30% to 90% fluctuating)
Now if I replace these 4 servers (HP DL320, 4GB RAM, dual core single Xeon processor) with 1 server (HP DL380,32GB RAM, quad core dual-Xeon processor), I can minimize the overall costing and improve overall performance as below:
1. Your rack U space (4U X1 = 4U in total)
2. Cooling power and operations (aKW X2 = 2aKW, a is considered as utilization per server)
3. Server power (bKW X2 = 2bKW, b is considered as utilization per server)
4. CPU utilization (50% to 90% fluctuating)
5. Memory utilization (30% to 90% fluctuating)
Thus it will save 50% of your power, will boost up you processes multiple times as you have plenty of RAM to utilize and dual processors, and the most important thing is saving your hardware cost upto 66%! And it's not like dream, one of the Warid data center and contact center has moved to Virtualization and they have saved 80% overall! like a dream but it's a true case story.
I don't know if I have killed your time, but you might be the correct person if you have decision taking ability over IT infrastructure costing, even for such a place like Bangladesh where our primary concern is the expenditure for the IT services. You can hold down the costing or hardware, power systems, cooling systems and rack space also simultaneously improve the performance better than anytime you have. You can read the case studies how Virtualization is being implementing worldwide and laying off costs, lots are available on google, more specifically at vmware, cisco, microsoft and oracle sites.
Regards,
Mijanur Rahman
CCIE Voice (wr), CCVP, CCNA
