Skip to content

DOC-1153 Add Datasize Estimates #968

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
priyankagunaki-cloud merged 4 commits into from
Aug 7, 2026
Merged

DOC-1153 Add Datasize Estimates #968

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
bf83761
DOC-1153 Add Datasize Estimates
Aug 4, 2026
a0020b3
DOC-1153 Update hw-and-sw-requirements.adoc
priyankagunaki-cloud Aug 5, 2026
01d2e8b
DOC-1153 Update hw-and-sw-requirements.adoc
priyankagunaki-cloud Aug 5, 2026
5e97954
Update hw-and-sw-requirements.adoc
priyankagunaki-cloud Aug 5, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
135 changes: 135 additions & 0 deletions modules/installation/pages/hw-and-sw-requirements.adoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,6 +65,7 @@ The software has been tested on the operating systems listed below:
| Windows (all version)
| X


| MacOS (Intel and M1 chip)
| X
|===
Expand Down Expand Up @@ -211,6 +212,140 @@ NOTE: Production systems should consider having replica clusters for redundancy
| 10 Gigabit Ethernet adapter
|===

=== Hardware Sizing Estimates by Dataset Size

The estimates in this section provide general guidance for sizing hardware based on dataset size, workload type, and compression ratio. They are intended to assist with initial capacity planning for a TigerGraph deployment.

Algorithmic workloads, such as PageRank, Community Detection, and Shortest Path, generally require more memory than non-algorithmic workloads because they perform additional graph computations.

@qingGongTG qingGongTG Aug 4, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

May just add the methods for computation as well.

Memory

Workload-type Memory sizing factor
Non-algorithmic 2 (1x for graph size + 1x for compute)
Algorithmic 3 (1x for graph size + 2x for compute)

Disk
Disk space per node = Roughly same as the memory per node.
If looking to store multiple backups, large data files, or export data locally, 2x the memory size per node is recommended.

@priyankagunaki-cloud priyankagunaki-cloud Aug 5, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@qingGongTG
I have committed the suggested changes. Please review


Compression ratio affects both memory and storage requirements. For example, a 10% compression ratio means the stored graph occupies approximately 90% of its original size, while a 40% compression ratio means it occupies approximately 60% of its original size.

==== Memory and Disk Sizing

The following workload-specific sizing factors are used to estimate memory requirements.

[cols="2,2",options="header"]
|===
|Workload type
|Memory sizing factor

|Non-algorithmic
|2 (1x for graph size + 1x for compute)

|Algorithmic
|3 (1x for graph size + 2x for compute)
|===

*Disk recommendations*

* Disk space per node should be approximately equal to the recommended memory per node.
* If the deployment stores multiple backups, large data files, or exported data locally, provision approximately 2× the recommended memory size for disk capacity.

==== Example

The following example demonstrates how workload type and compression ratio affect hardware sizing. The memory and disk estimates are calculated per node for the assumed 3-node HA cluster.

Assumptions:

* Raw dataset size: 500 GB
* HA cluster with 3 nodes
* Algorithmic workload
* CPU recommendation: Up to 64 cores

Using the sizing factors described above:

* At a 10% compression ratio:
** Memory: 1.35 TB
** Disk per node: ~2 TB

* At a 40% compression ratio:
** Memory: 900 GB
** Disk per node: ~1.35 TB


.Estimated hardware requirements per node by dataset size

[cols="1,1,1,1,1,1",options="header"]
|===
|Raw Data Size
|Compression Ratio
|CPU Recommendation
|Memory (Algorithmic)
|Memory (Non-Algorithmic)
|Disk per Node

@qingGongTG qingGongTG Aug 4, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

same, do you assume single node (this is total disk) or what?


|100 GB
|10%
|Up to 64 cores
|270 GB
|180 GB
|~405 GB

|100 GB
|40%
|Up to 64 cores
|180 GB
|120 GB
|~270 GB

|500 GB
|10%
|Up to 64 cores
|1.35 TB
|900 GB
|~2 TB

|500 GB
|40%
|Up to 64 cores
|900 GB
|600 GB
|~1.35 TB

|1 TB
|10%
|Up to 64 cores
|2.7 TB
|1.8 TB
|~4.05 TB

|1 TB
|40%
|Up to 64 cores
|1.8 TB
|1.2 TB
|~2.7 TB

|5 TB
|10%
|Up to 64 cores
|13.5 TB
|9 TB
|~20.25 TB

|5 TB
|40%
|Up to 64 cores
|9 TB
|6 TB
|~13.5 TB

|10 TB
|10%
|Up to 64 cores
|27 TB
|18 TB
|~40.5 TB

|10 TB
|40%
|Up to 64 cores
|18 TB
|12 TB
|~27 TB
|===

NOTE: These estimates are intended for initial capacity planning only. Actual hardware requirements vary depending on graph schema, workload characteristics, compression ratio, query complexity, replication factor, partitioning strategy, and performance objectives. For production deployments, consult TigerGraph Solution Architects to determine the appropriate hardware configuration.

=== Additional Considerations
This section describes considerations beyond sizing for the CPU, memory, and storage.
Expand Down