AI
Sunday, January 13, 2008 4:26:50 PM
Re: For this test Dell sent us a pre-production Barcelona server, based on their existing PowerEdge 2970. What it comes down to is Dell has done pretty much what we described on a previous page: switched out the dual core CPUs in an existing Opteron server for Barcelona models.
Interesting links, mas.
Given the power dissipation tests, while running Sunguard AA, the Barcelona system dissipates 301W of power, while the Stoakley system dissipated 364W of power.
However, under performance, the Barcelona system takes 366.9 seconds to complete the workload, while the Stoakley system takes 220.7 seconds.
Let's take a real world example. A company needs to process 10,000 units of this Sunguard AA application per day. How many servers of either type would they need?
1. there are 86,400 seconds in a day.
2. A single Stoakley system can process 10,000 units in 2,207,000 seconds; hence, it would take 26 servers to distribute those units to complete the work order in one day.
3. A single Barcelona system can process 10,000 units in 3,669,000 seconds; hence, it would take 43 servers to distribute those units to complete the work order in one day.
Now, here's the punchline:
4. 26 Stoakley systems will each dissipate 364W of power; hence, the total electricity for the server installation (just the boxes) would be 9.5kW.
5. 43 Barcelona systems will each dissipate 301W of power; hence, the total electricity for the server installation (just the boxes) would be 13.0kW.
A customer needing enough servers for that level of processing would have to account for more than 35% more power requirements using Barcelona for the same workload, as opposed to using Stoakley.
Hence, while single system Barcelona systems do in fact dissipate less power, the performance of the Stoakley system more than makes up for the performance/watt equation, which is more important to the end user than single system calculations.
Interesting links, mas.
Given the power dissipation tests, while running Sunguard AA, the Barcelona system dissipates 301W of power, while the Stoakley system dissipated 364W of power.
However, under performance, the Barcelona system takes 366.9 seconds to complete the workload, while the Stoakley system takes 220.7 seconds.
Let's take a real world example. A company needs to process 10,000 units of this Sunguard AA application per day. How many servers of either type would they need?
1. there are 86,400 seconds in a day.
2. A single Stoakley system can process 10,000 units in 2,207,000 seconds; hence, it would take 26 servers to distribute those units to complete the work order in one day.
3. A single Barcelona system can process 10,000 units in 3,669,000 seconds; hence, it would take 43 servers to distribute those units to complete the work order in one day.
Now, here's the punchline:
4. 26 Stoakley systems will each dissipate 364W of power; hence, the total electricity for the server installation (just the boxes) would be 9.5kW.
5. 43 Barcelona systems will each dissipate 301W of power; hence, the total electricity for the server installation (just the boxes) would be 13.0kW.
A customer needing enough servers for that level of processing would have to account for more than 35% more power requirements using Barcelona for the same workload, as opposed to using Stoakley.
Hence, while single system Barcelona systems do in fact dissipate less power, the performance of the Stoakley system more than makes up for the performance/watt equation, which is more important to the end user than single system calculations.
