Production Checklist

FAQ 1

Question:

How can I help make sure my TigerGraph instance is Production ready?

Answer Overview:

The following is a checklist intended for setting up a TigerGraph Production instance (either single node or cluster), it will include suggestions as well as best practices.

Please note this checklist is designed for TigerGraph version equal to or greater than 3.x

Disaster Recovery (please refer to the table below for a high-level comparison between GBAR and Cross-Region Replication)

Via GBAR (highest RPO and RTO): Backup and Restore - TigerGraph Server

Via Cross-Region-Replication (lowest RPO and RTO): Cross-Region Replication - TigerGraph Server

Architecture image

Backups

Automated scheduled (daily, hourly) backups (e.g. with cronjob)

Retention policy for disk space-saving (e.g. 6month is a common retention policy)

High-Availability

Recommended at least 3 replicas for each node

Each replica can be used for READ and WRITE transactions, it increases the concurrency

Minimize downtime

Machine specification:

8 core minimum (you might need more depending on your use case, data size, read/write ratio. Performance testing will clarify that)

CPU clock speed at least 3.2GHz (the higher the better)

Fast SSDs with IOPS > 3000

Optimal network bandwidth between your TigerGraph nodes

Misc

Monitoring (e.g. grafana) with paging alerts

Ideally dedicated machines only for TigerGraph (no other apps or processes running on it)

Load balancer between your app and TigerGraph