10 iSCSI Network Best Practices

iSCSI is a popular storage networking protocol that enables organizations to access storage resources over a network. It is a cost-effective and efficient way to access storage resources, but it is important to follow best practices when setting up an iSCSI network.

In this article, we will discuss 10 best practices for setting up an iSCSI network. Following these best practices will ensure that your iSCSI network is secure, reliable, and efficient.

1. Use a dedicated network for iSCSI traffic

When you use a dedicated network for iSCSI traffic, it ensures that the data is not competing with other types of traffic on the same network. This helps to ensure that your iSCSI traffic has enough bandwidth and low latency so that it can perform optimally. Additionally, using a dedicated network also helps to improve security since there are fewer potential attack vectors.

2. Use Jumbo Frames (MTU 9000)

Jumbo Frames are Ethernet frames that have a payload size of more than 1500 bytes. By using Jumbo Frames, you can increase the performance of your iSCSI network by reducing the number of packets sent over the wire and thus increasing throughput.

Jumbo Frames also reduce latency because they require fewer acknowledgements from the receiving end. This is especially important for applications such as video streaming or virtualization where low latency is essential.

To enable Jumbo Frames on an iSCSI network, make sure all devices in the network support it and configure them to use the same MTU value (e.g., 9000). Additionally, ensure that any switches used in the network also support Jumbo Frames and are configured accordingly.

3. Enable flow control on your switches and NICs

Flow control is a mechanism that helps prevent data loss due to congestion on the network. When enabled, it allows devices to pause and resume transmission of packets when needed.

Without flow control, if too much traffic is sent over the network at once, some of the packets may be dropped or corrupted. This can lead to poor performance and even data loss. By enabling flow control, you ensure that your iSCSI network runs smoothly and efficiently.

4. Disable Spanning Tree Protocol

Spanning Tree Protocol (STP) is a network protocol that helps prevent loops in the network. It works by blocking redundant paths and creating a loop-free topology for data transmission.

However, STP can cause latency issues on iSCSI networks due to its slow convergence time. This means that it takes longer for STP to detect changes in the network and adjust accordingly. As a result, disabling STP can help reduce latency and improve performance on an iSCSI network.

5. Configure the network switch to use multiple VLANs

Using multiple VLANs helps to isolate traffic and prevent congestion on the network. This is especially important for iSCSI networks, as they are typically used in high-performance environments where latency can be an issue. By using multiple VLANs, you can ensure that only the necessary traffic is sent over each VLAN, reducing the amount of unnecessary traffic and improving performance.

Additionally, configuring the switch to use multiple VLANs also provides a layer of security by isolating different types of traffic from one another. For example, if you have a web server and a database server connected to the same switch, you can configure two separate VLANs so that the web server’s traffic does not interfere with the database server’s traffic.

6. Set up redundant paths between your ESXi hosts and storage devices

Having redundant paths ensures that if one of the paths fails, your ESXi hosts can still access the storage devices. This is especially important in a production environment where downtime needs to be minimized. Redundant paths also provide better performance and reliability since data can be sent over multiple paths at once.

To set up redundant paths, you’ll need two physical network adapters on each ESXi host and two physical connections between the ESXi hosts and the storage device. You should then configure multipathing software such as VMware’s Native Multipathing Plugin (NMP) or EMC PowerPath/VE to ensure that both paths are used for communication.

7. Avoid using software initiators in production environments

Software initiators are not as reliable or secure as hardware-based initiators, and they can be more prone to errors.

Software initiators also require additional resources from the host system, which can lead to performance issues. Additionally, software initiators may not support all of the features available with a hardware-based initiator, such as booting from an iSCSI target.

For these reasons, it is best practice to use hardware-based initiators in production environments. This will ensure that your iSCSI network is reliable, secure, and performs optimally.

8. Use hardware-based initiators whenever possible

Hardware-based initiators are more reliable than software-based ones, as they don’t rely on the operating system or other applications to function properly. This means that if something goes wrong with the OS or an application, your iSCSI network won’t be affected. Additionally, hardware-based initiators can provide better performance and scalability than their software counterparts.

Finally, using a hardware-based initiator is often more cost effective in the long run, as it eliminates the need for additional software licenses and maintenance costs associated with software-based solutions.

9. Ensure that all of your ESXi hosts are configured with the same MTU size

When an ESXi host sends a packet over the iSCSI network, it needs to be able to receive a response from the target. If the MTU size of the source and destination hosts are different, then the packet will not be received correctly. This can lead to data corruption or even complete loss of connectivity.

By ensuring that all of your ESXi hosts have the same MTU size, you can avoid this issue and ensure reliable communication between the source and destination hosts. Additionally, setting the MTU size to 9000 bytes is recommended for optimal performance.

10. Verify that your physical network adapters support TCP Checksum Offload (TCO), Large Receive Offload (LRO), and jumbo frames

TCO and LRO are both features that help reduce the amount of processing power required to process network traffic. This helps improve overall performance, as it reduces the load on your CPU. Jumbo frames allow for larger packets of data to be sent over the network, which can also increase performance.

By verifying that your physical network adapters support these features, you can ensure that your iSCSI network is running at peak efficiency.