Pro Cluster - Fixed Storage
Last updated:2025-12-24 14:51:17
This document describes the specifications of TDMQ for Apache Pulsar pro clusters (fixed storage), including TPS, bandwidth, and the maximum number of topics, for reference during product selection.
Note:
1. The storage system of pro clusters uses three replicas to ensure high data availability.
2. Pro clusters support online specification upgrades. For specific operations, see Upgrading Cluster Specifications. The upgrade process is theoretically imperceptible to businesses. However, clients may experience disconnections and reconnections when the server rebalances topic traffic.
Cluster Specifications
No. | Cluster Specification Code | Maximum TPS for Message Sending and Receiving | Peak Bandwidth (MB/s) | Maximum Number of Topics Per Cluster | Minimum Storage Specifications (GB) | Maximum Number of Namespaces Per Cluster |
1 | PULSAR.P1.MINI2 | 2000 | 45 | 1000 | 200 | 50 |
2 | PULSAR.P1.SMALL4 | 4000 | 90 | 1000 | 200 | 50 |
3 | PULSAR.P1.SMALL6 | 6000 | 120 | 1000 | 267 | 50 |
4 | PULSAR.P1.SMALL10 | 10000 | 180 | 2000 | 400 | 100 |
5 | PULSAR.P1.MEDIUM15 | 15000 | 300 | 2000 | 333 | 100 |
6 | PULSAR.P1.MEDIUM20 | 20000 | 480 | 2000 | 400 | 100 |
7 | PULSAR.P1.MEDIUM40 | 40000 | 720 | 2000 | 533 | 100 |
8 | PULSAR.P1.MEDIUM60 | 60000 | 1080 | 2000 | 533 | 100 |
9 | PULSAR.P1.MEDIUM100 | 100000 | 1920 | 3000 | 533 | 150 |
10 | PULSAR.P1.LARGE150 | 150000 | 2400 | 3000 | 667 | 150 |
11 | PULSAR.P1.LARGE200 | 200000 | 3600 | 3000 | 800 | 150 |
12 | PULSAR.P1.LARGE400 | 400000 | 4200 | 3000 | 1000 | 150 |
13 | PULSAR.P1.LARGE600 | 600000 | 4800 | 3000 | 1200 | 150 |
14 | PULSAR.P1.LARGE1000 | 1000000 | 6000 | 3000 | 1667 | 150 |
Must-Knows
Maximum TPS for Message Sending and Receiving is calculated based on the sum of sent and received general messages with a message size of 4 KB. For messages with advanced features or large messages, the value should be multiplied by the corresponding multiplier. Peak Bandwidth indicates the maximum inbound and outbound bandwidth values. For detailed calculation methods, see Pro Cluster Pricing.
Storage Usage Instructions
1. For pro clusters (fixed storage), storage usage indicates the disk usage of the storage service bookie, not the storage space occupied by topic data in the broker.
2. The disk usage of the storage service bookie cannot be calculated based on the storage space occupied by all business topics in the broker. This is because the storage space occupied by topics constitutes only a portion of bookie storage, and additional space is consumed by index files, log files, and others.
3. Not only do business topic data consume the bookie disk space, but also cursors, schemas, and internal system topics occupy the bookie disk space.
4. Under the data reclamation mechanism of the bookie, after topic data is deleted from the broker, the bookie disk space may not be released immediately, resulting in write amplification. Generally, write amplification is within 30%, but can reach up to 50% in extreme scenarios. This is related to the data cleanup (called GC internally) mechanism of the bookie.
5. It is recommended that you configure and closely monitor cluster storage utilization alarms. In production environments, it is recommended that storage utilization be within 80% to prevent some nodes from reaching the full capacity in advance due to uneven storage distribution.
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback