# Client Diversity

## Importance of client diversity

Your Ethereum validator node is made up of various pieces of software clients working together to process and attest to transactions. Being a decentralised network, Ethereum is not dependent on any single physical server/node.

However, if every node runs only on one single client set, the existence of crazy bugs or targeted attacks on the software layer can still disrupt this process.

When this happens, we, as validator node operators, will suffer penalties. **These penalties increase exponentially when our validator nodes go offline together with a large portion of the network.**

Looking at the chart below, correlation penalties involving 33% or more of the network will result in all of your staked ETH being slashed!

<figure><img src="/files/40ieTTJDOK4AXRa4aspg" alt=""><figcaption><p>Kiln's research on correlation penalties</p></figcaption></figure>

This is why there are at least 4 - 5 clients to choose from for execution layer and consensus layer clients today. With Geth owning more than 50% of the execution layer client market share today, the existence of a crazy bug causing validator nodes running Geth to double-sign will be catastrophic.

This is why we will be recommending minority clients for this curriculum.

### Key considerations <a href="#key-considerations" id="key-considerations"></a>

Even though we should run minority clients, it does not automatically mean that any client can run on the same hardware. A key reason for Geth's popularity, other than its reliability, is the ability for it to run on minimal hardware.

**Execution layer clients (updated 4th Feb 2024)**

<table><thead><tr><th>Client</th><th width="118">CPU speed</th><th>RAM</th><th>Storage</th><th width="130">SSD speeds</th><th width="129">Sync speed</th><th>Market share</th></tr></thead><tbody><tr><td><strong>Geth</strong></td><td>lowest</td><td>16 GB</td><td>2 TB</td><td>Good</td><td>2nd</td><td>78%</td></tr><tr><td><strong>Besu</strong></td><td>lower</td><td>32 GB</td><td>2 TB</td><td>Good</td><td>1st</td><td>5%</td></tr><tr><td><strong>Nethermind</strong></td><td>lower</td><td>32 GB</td><td>2 TB</td><td>Good</td><td>3rd</td><td>14%</td></tr><tr><td><strong>Erigon</strong></td><td>higher</td><td>32 GB</td><td>4 TB*</td><td>Good</td><td>4th</td><td>2%</td></tr></tbody></table>

*\*Erigon is optimised for running archival nodes (full nodes are sufficient for validators) and will consume 2 TB of disk space within several weeks after syncing.*

*Source on market share -* [*https://clientdiversity.org/*](https://clientdiversity.org/)

**Consensus layer clients (updated 4th Feb 2024)**

Because the dependencies on hardware requirements is greater for the execution layer (EL) clients, the considerations for consensus layer (CL) clients will primarily be on market share.

| Client         | Market share |
| -------------- | ------------ |
| **Prysm**      | 38.93%       |
| **Lighthouse** | 33.01%       |
| **Teku**       | 18.69%       |
| **Nimbus**     | 8.21%        |
| **Lodestar**   | 1.16%        |

*Source on market share -* [*https://clientdiversity.org/*](https://clientdiversity.org/)

Further, because all CL clients now have the checkpoint sync feature - enabling CL syncing in just a few minutes - we can rapidly switch between CL clients with minimal downtime when there are bugs in one.

### Curriculum Scope <a href="#curriculum-scope" id="curriculum-scope"></a>

Because of the Key Considerations above, we will be covering the following clients as the scope for this DVT Home Staker Programme:

1. Execution layer: Nethermind and Besu
2. Consensus layer: Teku, Nimbus, Lodestar


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://nodebridge-africa.gitbook.io/nodebridge-africa/client-diversity.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.