76 lines
6.1 KiB
Markdown
76 lines
6.1 KiB
Markdown
# Reference architecture: up to 5,000 users
|
|
|
|
This page describes GitLab reference architecture for up to 5,000 users.
|
|
For a full list of reference architectures, see
|
|
[Available reference architectures](index.md#available-reference-architectures).
|
|
|
|
> - **Supported users (approximate):** 5,000
|
|
> - **High Availability:** True
|
|
> - **Test RPS rates:** API: 100 RPS, Web: 10 RPS, Git: 10 RPS
|
|
|
|
| Service | Nodes | Configuration ([8](#footnotes)) | GCP | AWS ([9](#footnotes)) | Azure([9](#footnotes)) |
|
|
|--------------------------------------------------------------|-------|---------------------------------|---------------|-----------------------|------------------------|
|
|
| GitLab Rails ([1](#footnotes)) | 3 | 16 vCPU, 14.4GB Memory | n1-highcpu-16 | c5.4xlarge | F16s v2 |
|
|
| PostgreSQL | 3 | 2 vCPU, 7.5GB Memory | n1-standard-2 | m5.large | D2s v3 |
|
|
| PgBouncer | 3 | 2 vCPU, 1.8GB Memory | n1-highcpu-2 | c5.large | F2s v2 |
|
|
| Gitaly ([2](#footnotes)) ([5](#footnotes)) ([7](#footnotes)) | X | 8 vCPU, 30GB Memory | n1-standard-8 | m5.2xlarge | D8s v3 |
|
|
| Redis ([3](#footnotes)) | 3 | 2 vCPU, 7.5GB Memory | n1-standard-2 | m5.large | D2s v3 |
|
|
| Consul + Sentinel ([3](#footnotes)) | 3 | 2 vCPU, 1.8GB Memory | n1-highcpu-2 | c5.large | F2s v2 |
|
|
| Sidekiq | 4 | 2 vCPU, 7.5GB Memory | n1-standard-2 | m5.large | D2s v3 |
|
|
| Object Storage ([4](#footnotes)) | - | - | - | - | - |
|
|
| NFS Server ([5](#footnotes)) ([7](#footnotes)) | 1 | 4 vCPU, 3.6GB Memory | n1-highcpu-4 | c5.xlarge | F4s v2 |
|
|
| Monitoring node | 1 | 2 vCPU, 1.8GB Memory | n1-highcpu-2 | c5.large | F2s v2 |
|
|
| External load balancing node ([6](#footnotes)) | 1 | 2 vCPU, 1.8GB Memory | n1-highcpu-2 | c5.large | F2s v2 |
|
|
| Internal load balancing node ([6](#footnotes)) | 1 | 2 vCPU, 1.8GB Memory | n1-highcpu-2 | c5.large | F2s v2 |
|
|
|
|
## Footnotes
|
|
|
|
1. In our architectures we run each GitLab Rails node using the Puma webserver
|
|
and have its number of workers set to 90% of available CPUs along with four threads. For
|
|
nodes that are running Rails with other components the worker value should be reduced
|
|
accordingly where we've found 50% achieves a good balance but this is dependent
|
|
on workload.
|
|
|
|
1. Gitaly node requirements are dependent on customer data, specifically the number of
|
|
projects and their sizes. We recommend two nodes as an absolute minimum for HA environments
|
|
and at least four nodes should be used when supporting 50,000 or more users.
|
|
We also recommend that each Gitaly node should store no more than 5TB of data
|
|
and have the number of [`gitaly-ruby` workers](../gitaly/index.md#gitaly-ruby)
|
|
set to 20% of available CPUs. Additional nodes should be considered in conjunction
|
|
with a review of expected data size and spread based on the recommendations above.
|
|
|
|
1. Recommended Redis setup differs depending on the size of the architecture.
|
|
For smaller architectures (less than 3,000 users) a single instance should suffice.
|
|
For medium sized installs (3,000 - 5,000) we suggest one Redis cluster for all
|
|
classes and that Redis Sentinel is hosted alongside Consul.
|
|
For larger architectures (10,000 users or more) we suggest running a separate
|
|
[Redis Cluster](../high_availability/redis.md#running-multiple-redis-clusters) for the Cache class
|
|
and another for the Queues and Shared State classes respectively. We also recommend
|
|
that you run the Redis Sentinel clusters separately for each Redis Cluster.
|
|
|
|
1. For data objects such as LFS, Uploads, Artifacts, etc. We recommend an [Object Storage service](../object_storage.md)
|
|
over NFS where possible, due to better performance and availability.
|
|
|
|
1. NFS can be used as an alternative for both repository data (replacing Gitaly) and
|
|
object storage but this isn't typically recommended for performance reasons. Note however it is required for
|
|
[GitLab Pages](https://gitlab.com/gitlab-org/gitlab-pages/issues/196).
|
|
|
|
1. Our architectures have been tested and validated with [HAProxy](https://www.haproxy.org/)
|
|
as the load balancer. Although other load balancers with similar feature sets
|
|
could also be used, those load balancers have not been validated.
|
|
|
|
1. We strongly recommend that any Gitaly or NFS nodes be set up with SSD disks over
|
|
HDD with a throughput of at least 8,000 IOPS for read operations and 2,000 IOPS for write
|
|
as these components have heavy I/O. These IOPS values are recommended only as a starter
|
|
as with time they may be adjusted higher or lower depending on the scale of your
|
|
environment's workload. If you're running the environment on a Cloud provider
|
|
you may need to refer to their documentation on how configure IOPS correctly.
|
|
|
|
1. The architectures were built and tested with the [Intel Xeon E5 v3 (Haswell)](https://cloud.google.com/compute/docs/cpu-platforms)
|
|
CPU platform on GCP. On different hardware you may find that adjustments, either lower
|
|
or higher, are required for your CPU or Node counts accordingly. For more information, a
|
|
[Sysbench](https://github.com/akopytov/sysbench) benchmark of the CPU can be found
|
|
[here](https://gitlab.com/gitlab-org/quality/performance/-/wikis/Reference-Architectures/GCP-CPU-Benchmarks).
|
|
|
|
1. AWS-equivalent and Azure-equivalent configurations are rough suggestions
|
|
and may change in the future. They have not yet been tested and validated.
|