Microservices optimizations
Introduction
REGARDS provides several type of configurations in order to optimize its performances:
- the number of parallel processes that can be run by microservice - this global thread pool will be shared by all
tenants on every instance of microservice (
regards.jobs.pool.size) - the number of parallel database connections permitted by tenant and by microservice - a connection pool is created
for every tenant on every instance of microservice (
regards.jpa.multitenant.maxPoolSize) - the number of entities handled inside a microservice process. Every REGARDS process exposes a property to override the number of entities handled on a bulk
Important
Usually, it is more efficient to configure each service to allow:
- a large number of parallel processes
- a small number of entities to process on every parallel process
But it depends on how many resources you can allocate to REGARDS services.
Implications
When you increase the number of parallel processes:
- you need to increase database connections
- you use more CPU,or vCPU, of your server
When you add more entities to handle by job/thread:
- you increase the amount of memory used on the server
- you increase the duration of the database commit
Default microservice configurations:
| Microservice | Default RAM | Default job pool size (shared) | Default DB pool size (by tenant) |
|---|---|---|---|
| rs-config | 128m | None | None |
| rs-registry | 128m | None | None |
| rs-admin | 512m | None | 5 per tenant of the instance |
| rs-admin-instance | 512m | None | 5 per instance |
| rs-access-project | 512m | None | 5 per tenant of the instance |
| rs-access-instance | 512m | None | 3 per instance |
| rs-dam | 1280m | None | 5 per tenant of the instance |
| rs-ingest | 1536m | 5 | 5 per tenant of the instance |
| rs-fem | 1024m | 5 | 5 per tenant of the instance |
| rs-notifier | 1280m | 5 | 10 per tenant of the instance |
| rs-storage | 1024m | 5 | 5 per tenant of the instance |
| rs-dataprovider | 1024m | 5 | 5 per tenant of the instance |
| rs-catalog | 1024m | 5 | 5 per tenant of the instance |
| rs-order | 1280m | 5 | 5 per tenant of the instance |
| rs-worker-manager | 1024m | 5 | 5 per tenant of the instance |
| rs-processing | 2048m | 50 | 5 per tenant of the instance |
| rs-lta-manager | 512m | 5 | 5 per tenant of the instance |
Many CPU adaptation
If your server(s) has a lot of CPU, or vCPU, we recommend you to increase on each microservice the value of job pool
size and DB pool size.
Don't forget to increase these properties together to avoid DB connection starvation during microservice processes.
Ingest optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.ingest.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of ingestion processes (ingest, delete, update...) | 2 |
Ansible property: group_docker_mservices.ingest.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.ingest.configuration.read_batch_size REGARDS property: regards.storage.client.response.batch.size | Number of storage request response to handle at a time | 500 |
Ansible property: group_docker_mservices.ingest.configuration.entities_per_job REGARDS property: regards.ingest.aips.scan.iteration-limit | Number of entities to handle in one ingestion process (ingest, delete, update...) | 100 |
FEM optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.fem.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of FEM processes (ingest, delete, update...) | 2 |
Ansible property: group_docker_mservices.fem.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.fem.configuration.read_batch_size REGARDS property: regards.fem.batch.size | Number of FEM requests to handle at a time | 1000 |
Ansible property: group_docker_mservices.fem.configuration.entities_per_job REGARDS property: regards.fem.max.bulk.size | Number of entities to handle in one FEM process (create, delete, update...) | 1000 |
Notifier optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.notifier.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of Notifier processes (ingest, delete, update...) | 2 |
Ansible property: group_docker_mservices.notifier.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.notifier.configuration.read_batch_size REGARDS property: regards.notifier.batch.size | Number of AMQP Notifier requests to handle at a time | 1000 |
Ansible property: group_docker_mservices.fem.configuration.entities_per_job REGARDS property: regards.notifier.max.bulk.size | Number of entities to handle in one Notifier process | 1000 |
Storage optimization
Note : It is recommended to not set a big value to the read_batch_size property for storage microservice.
Recommended value between 10 and 100.
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.storage.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of Storage processes (store, delete, copy...) | 2 |
Ansible property: group_docker_mservices.storage.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.storage.configuration.read_batch_size REGARDS property: regards.storage.availability.items.bulk.size | Number of file availibility requests to handle at a time | 10 |
Ansible property: group_docker_mservices.storage.configuration.read_batch_size REGARDS property: regards.storage.copy.items.bulk.size | Number of copy requests to handle at a time | 10 |
Ansible property: group_docker_mservices.storage.configuration.read_batch_size REGARDS property: regards.storage.deletion.items.bulk.size | Number of deletion requests to handle at a time | 10 |
Ansible property: group_docker_mservices.storage.configuration.read_batch_size REGARDS property: regards.storage.reference.items.bulk.size | Number of file reference requests to handle at a time | 10 |
Ansible property: group_docker_mservices.storage.configuration.read_batch_size REGARDS property: regards.storage.store.items.bulk.size | Number of store requests to handle at a time | 10 |
Ansible property: group_docker_mservices.storage.configuration.entities_per_job REGARDS property: regards.storage.cache.requests.per.job | Number of file availibility requests to handle in one Storage process | 100 |
Ansible property: group_docker_mservices.storage.configuration.entities_per_job REGARDS property: regards.storage.deletion.requests.per.job | Number of deletion requests to handle in one Storage process | 100 |
Ansible property: group_docker_mservices.storage.configuration.entities_per_job REGARDS property: regards.storage.storage.requests.per.job | Number of files to store requests to handle in one Storage process | 100 |
Ansible property: group_docker_mservices.storage.configuration.restoration_availability_bulk_limit REGARDS property: regards.storage.availability.request.product.bulk.limit | Maximum number of products inside a single availibility request. This limitation exists to avoid too many access to the storage location at a time. | 100 |
Data Management optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.dam.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.dam.configuration.crawler_batch_size REGARDS property: regards.crawler.max.bulk.size | Number of entities to request at a time from the data source to harvest | 10000 |
Ansible property: group_docker_mservices.dam.configuration.crawler_rate_delay_ms REGARDS property: regards.ingester.rate.ms | Time elapsed in milliseconds between two scans of configured crawlers to check if a datasource can be harvest | 60000 |
Order optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.order.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of order processes | 5 |
Ansible property: group_docker_mservices.order.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.order.configuration.entities_per_job REGARDS property: regards.order.batch.size | Number of entities to handle in one Order process | 1000 |
Dataprovider optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.dataprovider.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of Dataprovider processes | 5 |
Ansible property: group_docker_mservices.dataprovider.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.dataprovider.configuration.entities_per_job REGARDS property: regards.acquisition.batch.size | Number of entities to handle in one Dataprovider process | 100 |
Worker Manager optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.worker_manager.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.worker_manager.configuration.read_batch_size REGARDS property: regards.worker_manager.request.bulk.size | Number of Worker Manager requests to handle at a time | 1000 |
Processing optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.processing.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of Processing processes | 50 |
Ansible property: group_docker_mservices.processing.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
LTA Manager optimization
| Properties | Description | default |
|---|---|---|
Ansible property: group_docker_mservices.lta_manager.configuration.jobs_pool_size REGARDS property: regards.jobs.pool.size | Maximum number of LTA Manager processes | 5 |
Ansible property: group_docker_mservices.lta_manager.configuration.database_max_pool_size REGARDS property: regards.jpa.multitenant.maxPoolSize | Maximum number of parallel database connections usable | 5 |
Ansible property: group_docker_mservices.lta_manager.configuration.entities_per_job REGARDS property: regards.ltamanager.request.bulk.size | Number of entities to handle in one process (create, delete, update...) | 1000 |