> For the complete documentation index, see [llms.txt](https://mariadb.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://mariadb.com/docs/tools/enterprise/mariadb-enterprise-manager/usage/monitoring/dashboards/gridgain-8-cluster/caches.md).

# Caches

The GridGain 8 Caches dashboard in Enterprise Manager: per-cache throughput, size, transactions, latency, and hit and rollback ratios.

This dashboard shows per-cache performance for user data: throughput, size, transactions, and hit and rollback ratios.

{% hint style="info" %}
The panels show data only for caches with statistics enabled (`statisticsEnabled=true` in the cache configuration). A cache without statistics reports its counters as `0` and its size as `-1`.
{% endhint %}

### Overview

<figure><img src="https://2850516351-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkuTXWg0NDbRx6XUeYpGD%2Fuploads%2Fgit-blob-c9824db50f5fbea35da7452af173dff37d15547f%2Fem-gg8-caches-overview.png?alt=media" alt="Caches dashboard overview tiles"><figcaption></figcaption></figure>

| Panel                             | Description                                                                 |
| --------------------------------- | --------------------------------------------------------------------------- |
| **User Caches (reporting stats)** | Number of caches reporting statistics, including the internal system cache. |
| **Total Gets/s**                  | Get operations per second across all caches and nodes.                      |
| **Total Puts/s**                  | Put operations per second across all caches and nodes.                      |
| **Cluster Hit Ratio**             | Hits as a share of hits and misses. A low value means most reads miss.      |

### Cache Throughput

Operation rates with one series per cache on each node, so an imbalance between caches or a single node taking most of the traffic is visible.

<figure><img src="https://2850516351-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkuTXWg0NDbRx6XUeYpGD%2Fuploads%2Fgit-blob-936709aa39bfc793c2bba4b8a2325d244d3b0931%2Fem-gg8-caches-cache-throughput.png?alt=media" alt="Cache throughput section"><figcaption></figcaption></figure>

| Panel                                         | Description                                                                                          |
| --------------------------------------------- | ---------------------------------------------------------------------------------------------------- |
| **Cache Gets/s (all caches, all nodes)**      | Get rate per cache, per node.                                                                        |
| **Cache Puts/s (all caches, all nodes)**      | Put rate per cache, per node.                                                                        |
| **Cache Removals & Evictions/s (all caches)** | Removals against evictions. Removals come from the application; evictions come from memory pressure. |

### Cache Size & Off-Heap

<figure><img src="https://2850516351-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkuTXWg0NDbRx6XUeYpGD%2Fuploads%2Fgit-blob-4c9a9746310dacce36cbbbcdbb33192a77f5a62c%2Fem-gg8-caches-cache-size-off-heap.png?alt=media" alt="Cache size and off-heap section"><figcaption></figcaption></figure>

| Panel                                                | Description                                                                                                         |
| ---------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------- |
| **Cache Size — Entries (cluster sum per cache)**     | Entries per cache, including primary and backup copies. A sudden drop indicates eviction or a lost node.            |
| **Heap vs Off-Heap Entries (cluster sum per cache)** | Entries held on heap against entries held off heap.                                                                 |
| **Off-Heap Gets: Hits vs Misses/s**                  | Off-heap hits and misses per second. A miss means the entry had to come from the backing store or the primary node. |

### Cache Transactions

Average latencies are calculated as total operation time divided by the number of operations since node start. Caches with no operations show gaps.

<figure><img src="https://2850516351-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkuTXWg0NDbRx6XUeYpGD%2Fuploads%2Fgit-blob-a9781c88c938098950656ec21b96c32449ee43ae%2Fem-gg8-caches-cache-transactions.png?alt=media" alt="Cache transactions section"><figcaption></figcaption></figure>

| Panel                                                  | Description                                                                                                |
| ------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------- |
| **TX Commits & Rollbacks/s (all caches)**              | Commits against rollbacks per cache. A high share of rollbacks indicates contention or application errors. |
| **Get / Put / Remove Average Latency (per cache)**     | Average get, put, and remove time per cache since node start.                                              |
| **Commit & Rollback Average Latency (per cache)**      | Average commit and rollback time per cache since node start.                                               |
| **Write / Read Ratio (cluster-wide, all user caches)** | The mix of reads and writes in the workload.                                                               |

### Cache Hit / Miss & Transaction Ratios

How many transactions roll back and how many reads are served, independent of load. The ratios are cluster-wide totals across all user caches.

<figure><img src="https://2850516351-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FkuTXWg0NDbRx6XUeYpGD%2Fuploads%2Fgit-blob-8a1ffd7c5a4bdd14eb6d892d874bb6786b83aa06%2Fem-gg8-caches-cache-hit-miss-transaction-ratios.png?alt=media" alt="Cache hit, miss and transaction ratios section"><figcaption></figcaption></figure>

| Panel                                                           | Description                                                                    |
| --------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| **Rollback Ratio & Miss Ratio (cluster-wide, all user caches)** | Rollbacks as a share of all transactions, and misses as a share of all reads.  |
| **Hits vs Misses/s (per cache, per node)**                      | Hits and misses per cache, per node, to find the cache behind a low hit ratio. |

<sub>*This page is: Copyright © 2026 MariaDB. All rights reserved.*</sub>

{% @marketo/form formId="4316" %}
