Sizing and scalability considerations
The sizing baselines specified are based on the performance lab benchmark test results performed in BMC’s test labs. You can use these baselines for your on-premises BMC Helix IT Service Management deployment.
Important
To view the latest version of this topic, see
Sizing and scalability considerations
BMC’s performance testing is based on four different system usage profiles: compact, small, medium and large.
Small, medium, and large categories are suitable for production deployments and are based on the number of concurrent users.
Compact is a special sizing that is the minimum requirements for a functional BMC Helix Innovation Suite system. Compact systems are suitable for POC, development, and QA systems where resilience and system performance under load is not a consideration.
Concurrent users indicates the combined number of users that are logged in and actively working on a system across each of the following BMC Helix IT Service Management applications that are hosted on a single deployment of BMC Helix IT Service Management 21.05:
- BMC Helix ITSM
- BMC Helix ITSM: Smart IT
- BMC Helix ITSM: Smart Reporting
- BMC Digital Workplace
- BMC Digital Workplace Catalog
- BMC Helix Business Workflows
- BMC Live Chat
For real-world deployments, concurrent users might not be the only factor that drives the categorization of a system in the small, medium, or large category.
For example, if you have a high number of integrations that create a high number of system transactions, you need to increase your sizing to accommodate the additional load from integrations.
| Category | Concurrent Users | Maximum Concurrent Users Per Application* | Definition | |
|---|---|---|---|---|
Smart IT | BMC Digital Workplace | |||
| Compact | NA | NA | NA | Minimum functional deployment. Suitable for POC, development, and QA systems. |
| Small | 730 | 150 | 100 | Suitable for production systems where overall load does not exceed the number of concurrent users. |
| Medium | 2110 | 550 | 225 | |
| Large | TBA | TBA | TBA | |
*Maximum Concurrent Users Per Application: Smart IT and BMC Digital Workplace are resource-intensive applications.
If the number of concurrent users for these applications exceeds the stated maximum, use the higher appropriate sizing. For example, if you have a total number of 600 concurrent users, use small sizing. However, if you have 200 concurrent Smart IT users, use medium sizing.
Other applications on the BMC Helix Innovation Suite platform do not have these restrictions if the total number of concurrent users does not exceed the stated maximums values.
Kubernetes infrastructure sizing requirements
The following table shows the compute requirements for Kubernetes deployments of the containerized BMC Helix IT Service Management 21.05 version. Compute requirements are the combined requirements of CPU, RAM, and Persistent Volume Disk requirements for the Kubernetes compute nodes.
| Category | CPU | RAM (GB) | Persistent Volume Disk (GB) |
|---|---|---|---|
Compact | 118 | 210 | 200 |
Small | 184 | 309 | 200 |
Medium | 189 | 434 | 200 |
Per worker node logging
Combine the overall Kubernetes compute nodes requirement with the following requirement for per worker node logging that runs as a
DaemonSet
on each worker node:
vCPU per worker node | RAM per worker node |
|---|---|
3.1 | 8.2 |
For example, for a compact deployment, calculate the overall compute footprint as follows:
| Compact deployment on | CPU | RAM |
|---|---|---|
| 5 worker nodes | 118 + (5 x 3.1) = 133.5 | 210 + (5 X 8.2) = 251 |
| 3 Worker Nodes | 118 + (3 x 3.1) = 127.3 | 210 + (3 X 8.2) = 243.6 |
Important
- Compact sizing is not appropriate for production systems that require redundancy and high availability because it uses a number of single replicas of critical services.
BMC Helix IT Service Management 21.3 includes additional components such as BMC Helix Dashboards and ITSM Insights that will require additional resources.
CPU requirements
The CPU requirements are as follows:
- CPU size must be minimum 2.4 Gz.
BMC expects future releases of BMC Helix IT Service Management will require CPUs that support Tensorflow 2.x.
Most of the modern chipsets support this capability. For more information, see Hardware requirements
in TensorFlow documentation.
Compute node minimum requirements
Each node in the Kubernetes cluster must have the following minimum configurations:
| Specification | Compute node |
|---|---|
CPU | 8 |
Memory | 32 GB |
Available system disk after OS installation | 150 GB |
For high availability deployments such as production environments, Kubernetes recommends three control nodes.
Best practice
We recommend that you review the vendor documentation for sizing of control nodes. BMC does not have any specific requirements for the sizing of control nodes in a Kubernetes cluster.
For information about recommendations for canonical Kubernetes control nodes, see
Size of master and master components
in Kubernetes documentation.
Persistent volume disk requirements
High performance of Kubernetes Persistent Volume Disk (PVC) is essential for the overall system performance. Persistent Volume Disk requires block storage. BMC supports a Bring-Your-Own-Storage model for Kubernetes Persistent Volumes. BMC lab testing uses Ceph storage.
We recommend that you use solid-state drive (SSD) with the following configurations:
| PVC | Recommendation |
|---|---|
Write latency | 1 ms |
Read latency | 1 ms |
| Write Throughput | 30 MBPS |
| Read Throughput | 80 MBPS |
| IOPS Write | 20 K |
| IOPS Read | 7 K |
Comments
Do we really really need so much computing power in order deploy helix itsm? Almost 200 CPUs, more than 300 GB of RAM... We have customers who is using version 20.02 on two servers with 32 GB or ram in each. How can we justify that if they want to upgrade 21.05 or any version above, they need to spare 5-10 times of their current hardware resource?
Someone from BMC Support told me that: "The HELIX deployment offers no flexibility in these requirements, if you don't meet at least the compact sizing requirements then you will not be able to deploy."
If that is the case... Houston, we have a problem!
Thanks for sharing your feedback Emrah.
Sizing is based on our performance benchmark test results performed in BMC’s test labs. Benchmarking is a challenging topic and our current guidance is not optimal.
We are closely looking at this topic and you can expect some improvements over the coming releases.
Please may I recommend joining our BMC Communities for Containers. Below is the Link. https://community.bmc.com/s/group/0F93n000000ILQYCA4/bmc-helix-containerization.
This is indeed a great forum for these questions, and where we expect to share our plans on future optimizations?
Log in or register to comment.