How much energy can I save using Flash in storage?
Many have heard that in the west, data center owners are constantly concerned about the amount of energy consumed and that their maintenance is very expensive. One of the most voracious components of the data center is the storage system , this is due to the fact that the latter use a large number of mechanical hard drives. Many have also heard that electricity "over the hill" is much more expensive. And what about us?
After reading a short article about comparing two NetApp FAS3250 and FAS8020 models , I thought, “How much cheaper is Flash in terms of power consumption if we compare these two configurations?”
Knowing the power consumption of storage systems and configurations that were taken from the spec.org site for testing storage performance , we can conclude that some systems, being more expensive at the initial purchase, can “lose a few extra pounds” and become cheaper compared to competitors in the cost of maintenance, simply because they use flash technology.
Below is the power consumption data for the following storage systems .
The FAS8020 is the “successor” of the older FAS3220, which in turn is “lower class” than the FAS3250.
Two tables were created:
The first one calculates the cost of electricity costs and is available for viewing .
The second table is for entering data and is editable .
The separation of the tables is done so that someone inadvertently does not delete or change the formulas.
The cost of electricity (cell i2) for Kiev can be found on the website of the National Energy Commission. I suggest taking the 4th item from the top, i.e. in excess of 800 kW * hour per month, which corresponds to 95.76 kopecks or 0.9576 UAH, which in turn is now approximately equal to $ 0.1. For versatility, the form uses US dollars. Of course, everyone can set the price indicated on his electricity bill.
Tables are designed so that you can compare any of your values.
We go to the table " for editing ".
We put in the cell B3 consumption for the first system, C3 - Worst case of consumption by the system.
We add the consumption for the second system to cell D3 , E3 - The worst case for the second system.
These values are indicated in units of kilowatt * hour .
Next, enter the price, in US dollars, in cell i2 .
We go to the table " to view"(here perhaps you need to wait a bit while google pulls up the data from the first table) and compare the difference between the power consumption of the first and second systems in the two Diff columns : Typical and Worst Case.
If suddenly someone accidentally erased everything from the table" for editing " , go to the tab Original Input Data (table for viewing ) and just copy everything into a table for editing. I want to draw your attention to the fact that Google for some reason does not always pull up data from other tables or not always on time or not always the most current data - here you need to either wait or reload the page in a new browser window. In addition, there may be a case when several people will work on editing the table at the same time, the number of simultaneously open sessions is visible in the upper right - use the comment function in the file to agree on the order of changes or just wait until everyone leaves.
As you can see from the FAS8020 performance test, it’s slightly better in bandwidth and slightly better in response speed (latensi), while the cost of electricity for 5 years in this system is approximately 300 thousand kilowatt hours per hour and therefore cheaper by about 30 $ thousand than those of similar systems used on lshim number of spindles. Both systems are available for order through NetApp channels (as of 03/18/2014).
Despite the fact that air conditioning costs are not included in this calculation, do not forget that as a rule the amount of energy consumed is almost equal to or slightly lower than the amount of heat generated. And this heat must be removed not by cheap powerful air conditioners, which also consume electricity.
Although this rule for saving money with Flash may not always work, nevertheless, this technology provides an additional “regulator” for choosing storage systems and allows improving performance indicators and response to reduce maintenance costs for both storage systems and the data center as a whole.
Please write comments on typos in the text and suggestions for its correction to the PM .
After reading a short article about comparing two NetApp FAS3250 and FAS8020 models , I thought, “How much cheaper is Flash in terms of power consumption if we compare these two configurations?”
Knowing the power consumption of storage systems and configurations that were taken from the spec.org site for testing storage performance , we can conclude that some systems, being more expensive at the initial purchase, can “lose a few extra pounds” and become cheaper compared to competitors in the cost of maintenance, simply because they use flash technology.
Below is the power consumption data for the following storage systems .
| Model & Config | Typical, kWatt * h | Worst Case, kWatt * h | Cell per |
|---|---|---|---|
| FAS3250 | 9.68 | 10.03 | B3 and C3 respectively |
| FAS8020 | 2,35 | 3,51 | D3 and E3 respectively |
The FAS8020 is the “successor” of the older FAS3220, which in turn is “lower class” than the FAS3250.
Two tables were created:
The first one calculates the cost of electricity costs and is available for viewing .
The second table is for entering data and is editable .
The separation of the tables is done so that someone inadvertently does not delete or change the formulas.
The cost of electricity (cell i2) for Kiev can be found on the website of the National Energy Commission. I suggest taking the 4th item from the top, i.e. in excess of 800 kW * hour per month, which corresponds to 95.76 kopecks or 0.9576 UAH, which in turn is now approximately equal to $ 0.1. For versatility, the form uses US dollars. Of course, everyone can set the price indicated on his electricity bill.
How to use tables
Tables are designed so that you can compare any of your values.
We go to the table " for editing ".
We put in the cell B3 consumption for the first system, C3 - Worst case of consumption by the system.
We add the consumption for the second system to cell D3 , E3 - The worst case for the second system.
These values are indicated in units of kilowatt * hour .
Next, enter the price, in US dollars, in cell i2 .
We go to the table " to view"(here perhaps you need to wait a bit while google pulls up the data from the first table) and compare the difference between the power consumption of the first and second systems in the two Diff columns : Typical and Worst Case.
If suddenly someone accidentally erased everything from the table" for editing " , go to the tab Original Input Data (table for viewing ) and just copy everything into a table for editing. I want to draw your attention to the fact that Google for some reason does not always pull up data from other tables or not always on time or not always the most current data - here you need to either wait or reload the page in a new browser window. In addition, there may be a case when several people will work on editing the table at the same time, the number of simultaneously open sessions is visible in the upper right - use the comment function in the file to agree on the order of changes or just wait until everyone leaves.
Saving with Flash
As you can see from the FAS8020 performance test, it’s slightly better in bandwidth and slightly better in response speed (latensi), while the cost of electricity for 5 years in this system is approximately 300 thousand kilowatt hours per hour and therefore cheaper by about 30 $ thousand than those of similar systems used on lshim number of spindles. Both systems are available for order through NetApp channels (as of 03/18/2014).
Cooling
Despite the fact that air conditioning costs are not included in this calculation, do not forget that as a rule the amount of energy consumed is almost equal to or slightly lower than the amount of heat generated. And this heat must be removed not by cheap powerful air conditioners, which also consume electricity.
Conclusion
Although this rule for saving money with Flash may not always work, nevertheless, this technology provides an additional “regulator” for choosing storage systems and allows improving performance indicators and response to reduce maintenance costs for both storage systems and the data center as a whole.
Please write comments on typos in the text and suggestions for its correction to the PM .