ABCDEFGHIJKLMNOPQRST
1
SQLFire Sample Capacity Planning WorksheetV03
2
3
Directions:To enter your own data or check formulas, go to File > Make a Copy. You will have your own version on your own drive.
4
Use the table below to get an estimated amount of storage space. The total storage space is used as input to the second table and set of calculations to show the number and size of servers.
5
Fill out the white cells. For each table in column C, fill out columns D, E, F, G, I and J. Gray columns are calculated.
6
"OTHER Tables" is used as a generic catchall. It is not intended to replace a row for an important table. Add more rows for working tables.
7
If you need to add rows for more tables, make sure you replicate the calculations in columns P thru S to the new rows. These calculations are used in column H and subsequent forumulas.
8
* For more information on row sizes, see the documentation: http://pubs.vmware.com/vfabric53/index.jsp?topic=/com.vmware.vfabric.sqlfire.1.1/data_management/memory_analytics/memory_requirements_chapter.html
9
10
Estimate Storage Space
11
Table NamesRaw data size per row*Estimated duplicate rows in tableOverflow tablePersistent tableDisk OverheadPrimary IndexNumber of secondary indexesIndex OverheadEstimated Number of RowsEstimated storage (GB)Table Used for Calculations.
Do not remove or change!
12
TABLE140960YN64Y31521,000,0004.0200064
13
TABLE240961NY120N001,000,0007.85001200
14
TABLE31503NN64N001,000,0000.8006400
15
TABLE41501YY152Y11041,000,0000.76152000
16
TABLE570564YY152Y21281,000,00034.16152000
17
TABLE61001YY152Y21281,000,0000.71152000
18
TABLE730720YY152Y21281,000,0003.12152000
19
TABLE82005YY152Y21281,000,0002.68152000
20
TABLE91000YY152Y21281,000,0000.35152000
21
OTHER Tables5.00
22
Totals9,000,00059.45
23
Directions:
24
In the table below, enter data into the white cells with red text.
25
Choose the number of days you plan to store data in memory.
26
Enter the JVM head room percentage and JVM heap size.
27
Enter the RAM per physical machine.
28
29
Estimate Number of Machines and RAMAbout these Calculations:
30
59.45GB of primary storage per dayThe table to the left helps to allocate the number of machines and RAM.
This brief narrative helps to explain how we apply JVMs and VMs to physical machines.

From the prior table, we take the daily storage as input. In this case, it is almost 60GB.

The number of days we store data in this example is 6, so we have almost 360 GB of
RAM needed. Then, we add 40% for JVM overhead of 143 GB and make the data
redundant to bring the amount to 1 TB ((360+143)*2).

We assume a decision to make JVM heap size at 32GB (for performance) and divide
the 1 TB by this amount to end up with 16 JVMs for primary RAM and the same for redundant.

Next, we allocate to physical machines. If you are hosting 32 GB JVMs on 352GB of RAM,
then you can have 352/32 JVMS or 11 JVMs. Before, we caluclated 16 JVMs for our primary
data so we would need 2 machines of the 384 GB RAM size.

Of course, the pieces of data can change for particular deployments.
31
6.00Number of days to store
32
356.69GB of primary data for 6 days
33
40%JVM Head Room
34
143.00Total Primary RAM required (for JVM overhead)
35
999.39Total Data w/redundancy (Data RAM + JVM RAM) *2
36
1000.00Total RAM Required w/Redundancy and JVM overhead
37
32Heap size per JVM
38
16PrimaryNodes required
39
16Redundant nodes required
40
512.00Actual RAM usage per machine (at 16 nodes)
41
384.00RAM per machine
42
32.00RAM per machine available for other processes
43
352.00RAM per machine available to SQLF
44
11.00Max Nodes per box (32GB per/node)
45
2.00Primary Machines needed
46
2.00Secondary Machines needed
47
4.00Physical boxes needed for setup
48
1536GB Total RAM Installed
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78