Difference between revisions of "Scalability testing"
| Line 1: | Line 1: | ||
[[Main Page]] > [[Development roadmap]] > Scalability testing | [[Main Page]] > [[Development roadmap]] > Scalability testing | ||
| + | |||
| + | = Current Plans = | ||
| + | January 11, 2013 - Artefactual is committing resources towards Scalability Testing for Archivematica over the coming weeks and months. | ||
| + | |||
| + | The main goals of this effort are: | ||
| + | |||
| + | * set up a dedicated testing environment | ||
| + | The testing environment will start with 5 virtual machines set up in a hosted environment, where hardware resources can be scaled up and down between tests. It is expected that the test environment will be ready to use by January 15th. | ||
| + | |||
| + | |||
| + | * develop a new initial repeatable test suite | ||
| + | Initial tests will focus on two main areas - file io and documenting individual micro-service performance. Data will be collected from external monitoring tools as well as from internal instrumentation. | ||
| + | |||
| + | External monitoring will be done with two open source packages, munin and collectd. This will provide data at the operating system level. | ||
| + | Internal instrumentation already exists within the Archivematica source code, where each step in the process has a start time and end time recorded in the local database. This instrumentation will be extended and refined during the buildout of the test suite. The data collected will be used to identify which specific micro-services, and which steps within those micro-services are taking the longest time to complete. | ||
| + | |||
| + | * document a full matrix of test parameters | ||
| + | Archivematica workflow can vary considerably depending on use case. Artefactual will document all testing efforts on this wiki, building out a matrix of test cases. For example, we expect that adding additional storage subsystem capacity will allow for linear growth in scalability (add more disks, it should all go faster). This will be one of the first 'columns' in our test matrix, repeating tests with the same workload, changing the capacity (maximum io's per second) of the storage subsystem between tests. | ||
| + | |||
| + | Initial tests will focus on the 4 primary stages in the Archivematica workflow - Transfer, Ingest, creation of SIP, creation of AIP. There are additional steps required, both before Transfer, and after creation of AIP, however these steps do not necessarily involve the use of Archivematica code. For example, moving digital objects to a shared folder that Archivematica can access is a prerequisite of the Transfer stage, and can take a considerable amount of time. We will document best practices for how to complete that work after initial scalability testing is complete. | ||
| + | |||
| + | * repeat test suite at customer sites | ||
| + | The two initial customer sites have been identified by Archivematica and tests will be repeasted at both customer sites. | ||
| + | |||
| + | |||
| + | |||
| + | = Test Structure = | ||
| + | |||
| + | Scalability testing is done using a scripted workload, where all decision points, that are normally left to the Archivist to make using the Dashboard, are instead automated through the use of a configuration file. This allows for repeatable test cases. Example test scripts will be posted here over the coming weeks. | ||
| + | |||
| + | |||
= Test File Sets = | = Test File Sets = | ||
Revision as of 11:29, 11 January 2013
Main Page > Development roadmap > Scalability testing
Current Plans
January 11, 2013 - Artefactual is committing resources towards Scalability Testing for Archivematica over the coming weeks and months.
The main goals of this effort are:
- set up a dedicated testing environment
The testing environment will start with 5 virtual machines set up in a hosted environment, where hardware resources can be scaled up and down between tests. It is expected that the test environment will be ready to use by January 15th.
- develop a new initial repeatable test suite
Initial tests will focus on two main areas - file io and documenting individual micro-service performance. Data will be collected from external monitoring tools as well as from internal instrumentation.
External monitoring will be done with two open source packages, munin and collectd. This will provide data at the operating system level. Internal instrumentation already exists within the Archivematica source code, where each step in the process has a start time and end time recorded in the local database. This instrumentation will be extended and refined during the buildout of the test suite. The data collected will be used to identify which specific micro-services, and which steps within those micro-services are taking the longest time to complete.
- document a full matrix of test parameters
Archivematica workflow can vary considerably depending on use case. Artefactual will document all testing efforts on this wiki, building out a matrix of test cases. For example, we expect that adding additional storage subsystem capacity will allow for linear growth in scalability (add more disks, it should all go faster). This will be one of the first 'columns' in our test matrix, repeating tests with the same workload, changing the capacity (maximum io's per second) of the storage subsystem between tests.
Initial tests will focus on the 4 primary stages in the Archivematica workflow - Transfer, Ingest, creation of SIP, creation of AIP. There are additional steps required, both before Transfer, and after creation of AIP, however these steps do not necessarily involve the use of Archivematica code. For example, moving digital objects to a shared folder that Archivematica can access is a prerequisite of the Transfer stage, and can take a considerable amount of time. We will document best practices for how to complete that work after initial scalability testing is complete.
- repeat test suite at customer sites
The two initial customer sites have been identified by Archivematica and tests will be repeasted at both customer sites.
Test Structure
Scalability testing is done using a scripted workload, where all decision points, that are normally left to the Archivist to make using the Dashboard, are instead automated through the use of a configuration file. This allows for repeatable test cases. Example test scripts will be posted here over the coming weeks.
Test File Sets
Test design
Maximums to test for:
- Max number of SIPS - 10
- Max number of files in SIP - 10,000
- Max size of individual file - 30 GiB
- Max size of SIP - 100 GiB
Baseline amounts:
- number of SIPS - 1
- number of files in SIP - 10
- size of individual file - 1 MiB
- size of SIP - 100 MiB
| Test | No. of SIPs | No. of files in SIP | Max size of individual file | Max size of SIP |
|---|---|---|---|---|
| 1. Baseline Test | 1 | 10 | 1 MiB | 100 MiB |
| 2. No. of SIPs | 10 | 10 | 1 MiB | 100 MiB |
| 3. No. of files | 1 | 10,000 | 1 MiB | 100 MiB |
| 4. Max file size | 1 | 10 | 30 GiB | 100 MiB |
| 5. Max SIP size | 1 | 10 | 1 MiB | 100 GiB |
| ... |
- Other tests: combination of maximums
CVA tests
System setup:
- Bare-metal install, 1 processor
- 2 cores
- 4GB ram 9 GB swap
- xubuntu
Note: excludes store AIP and upload DIP micro-services except where noted
| Test date | No. transfers/SIPs | No. files | Total file size | Largest file size | AIP size | Total time | Comments |
|---|---|---|---|---|---|---|---|
| 2011/11/10 | 1/1 | 1,000 | 12.1 GB | 60 MB |
| ||
| 2011/11/10 | 1/1 | 1 | 2.7 GB | 2.7 GB | Failed at prepareAIP due to max Bag size: | ||
| 2011/11/18 | 1/1 | 1,000 | 12.1 GB | 60 MB | 7.2 GB | 4 hrs 30 mins | Access normalization only |
| 2011/12/02 | 2/2 | 1,998 | 13 GB | 21 MB | Access normalization only | ||
| 2011/12/11 | 1/1 | 1,000 | 6.51 GB | 21 MB | 3.5 GB | Access normalization only | |
| 2011/12/11 | 2/2 | 1,996 | 13.8 GB | 27 MB | 7.2 GB | Access normalization only | |
| 2011/12/13 | 3/3 | 2,974 | 18.6 GB | 20 MB | 10.3 GB | 3 hrs 19 mins | Access normalization only |
| 2011/12/14 | 4/4 | 3,993 | 24.6 GB | 22 MB | 13.2 GB | 3 hrs 16 mins | Access normalization only |
| 2011/12/15 | 4/4 | 3,982 | 43 GB | 12 MB | 15 GB | 3 hrs 30 mins | Access normalization only |
| 2011/12/15 | 6/6 | 5,113 | 34.1 GB | 38 MB | 19.8 GB | 4 hrs 2 mins | Access normalization only |
| 2012/01/04 | 6/6 | 5,845 | 42.4 GB | 33 MB | 24 GB | 3 hrs 52 mins | Access normalization only |
| 2012/01/05 | 3/3 | 2,957 | 20.9 GB | 45 MB | 13.6 GB | 4 hrs | Access normalization only |
| 2012/01/05 | 6/6 | 5,947 | 33 GB | 52 MB | 19.2 GB | 4 hrs 47 mins | Access normalization only |
| 2012/01/12 | 6/6 | 4,847 | 38.5 GB | 58 MB | 23.2 GB | 4 hrs 43 mins | Access normalization only |
| 2012/01/13 | 6/6 | 5,912 | 101.6 GB | 175 MB | 63.8 GB | 8 hrs 53 mins | Access normalization only |
| 2012/01/17 | 1/1 | 1 | 1.4 GB | 1.4 GB | 0.6 GB | 25 mins | Access normalization only |
| 2012/01/17 | 5/5 | 23 | 19.7 GB | 2.1 GB | 19 GB | 4 hrs 1 min | Access normalization only |
| 2012/01/18 | 2/2 | 2 | 3.8 GB | 2.1 GB | 3.7 GB | 1 hr 11 mins | Access normalization only |
| 2012/01/20 | 6/6 | 14 | 6.1 GB | 1.3 GB | 5.9 GB | 48 mins | Access normalization only |
| 2012/02/07 | 5/5 | 5 | 56.7 GB | 25.4 GB | 55.5 GB | 4 hrs 51 mins | No normalization |
| 2012/02/08 | 5/5 | 10 | 124.4 GB | 23.8 GB | 122.2 GB | 8 hrs 21 mins | No normalization |
| 2012/02 | 1/1 | 1044 | 7.5 GB | 12.4 MB | 32.8 GB | >16 hrs | Preservation and access normalization |
| 2012/02 | 1/1 | 104 | 611.6 MB | 7.1 MB | 2.58 GB | <2 hrs | Preservation and access normalization |
| 2012/02 | 1/1 | 2125 | 47.1 GB | 35.9 MB | 46.2 GB | >24 hrs | Preservation and access normalization |
| 2012/03 | 1/1 | 1654 | 7.9 GB | 11.7 MB | 37.7 GB | >16 hrs | Preservation and access normalization |
| 2012/03 | 1/1 | 1195 | 5.7 GB | 9.9 MB | 26.8 GB | >12 hrs | Preservation and access normalization |
| 2012/03/22 | 1/1 | 11.0 GB | 246.3 MB | GB | Preservation and access normalization | ||
| 2012/03/22 | 1/1 | 6.7 GB | 9.7 MB | GB | Preservation and access normalization | ||
| 2012/03/26 | 1/1 | 6.6 GB | 14.3 MB | GB | Preservation and access normalization | ||
| 2012/03 | 1/1 | 18.1 GB | 11.7 MB | Preservation and access normalization |
Multi-processor testing
Problem statement
- Does the amount of processing time decrease for each additional processing station added?
- If yes, by how much?
Constants and variables
Constants:
- Ram amount
- Ram speed
- Disk size
- Cpu frequency
Variables:
- Number of clients
- Number of transfer(s)
- Size of transfer(s)
- Number of files(s)
Ideal network for testing network consists of 6nodes+ each with dual core processor, 2GB+ memory, and 6GB+ disk space. Due to limited disk capacity, current tests are running with 5 nodes.
Testing data
- All testing data will be be preserved for analysis. Select data will be reported on this wiki.
Network setup
| HOSTNAME | Processor | Memory | Disk/s Size | IP | Filesystem | Services | Network Connection Speed | Ram speed/timing | Shared directory disk write speed | Shared directory disk read speed |
|---|---|---|---|---|---|---|---|---|---|---|
| test01server | 4x500mhz | 2048mb | 6GB+35GB | 10.10.0.1 | ext4 | MCPServer,MySQL,NFS,MCPClient | ||||
| test01client01 | 2x500mhz | 1024mb | 6GB | 10.10.0.11 | ext4,NFS | MCPClient | ||||
| test01client02 | 2x500mhz | 1024mb | 6GB | 10.10.0.12 | ext4,NFS | MCPClient | ||||
| test01client03 | 2x500mhz | 1024mb | 6GB | 10.10.0.12 | ext4,NFS | MCPClient | ||||
| test01client04 | 2x500mhz | 1024mb | 6GB | 10.10.0.14 | ext4,NFS | MCPClient |
Testing metrics
Our results are derived from running 000.zip through the archivematica pipeline using standard processing configuration settings, and then extracting MYSQL- timing views from the database. This gives us a clearer picture of productivity of clients.
two scripts are used to extract testing data from the database:
- automatedDistributedTestingReports.sh
- automatedDistributedTestingProcessingMachineInformationGathering.sh
After you have run your test data through archivematica they are to be used:
./automatedDistributedTestingReports.sh ./automatedDistributedTestingProcessingMachineInformationGathering.sh
you will recieve a similar fileset to this
2012.05.02-11.52.12_server_jobDurationsView.html 2012.05.02-11.52.12_server_MCP_DUMP.sql 2012.05.02-11.52.12_server_mysql_status.log 2012.05.02-11.52.12_server_netstat_summary.log 2012.05.02-11.52.12_server_PDI_by_unit.html 2012.05.02-11.52.12_server_processingDurationInformation.html server_2012.05.02-11.52.05_cpuinfo.log server_2012.05.02-11.52.05_free.log server_2012.05.02-11.52.05_IP.log
Test results
- Ram amount =
- Ram speed =
- Disk size =
- CPU frequency =
- Number of transfers =
- Total number of files =
- Total transfer size =
| No. of processors | Total processing time | Longest job | Second longest job | Third longest job |
|---|---|---|---|---|
| 1 | ||||
| 2 | ||||
| 6 |