Ever since installing SG500X switches in my network, i have had issues with database disconnects. I have 2 SG500X stacked together. I have many LAG groups spanning the 2 switches for redundancy. I recently installed an iSCSI SAN for 2 physical vmware hosts, and the latency between the hosts and the SAN makes it near unusable.
When looking at the switches, everything seems fine. When looking at the vmware cli, you can see all the latency occurring throughout the day. It shows significant latency for both vmware boxes over the iSCSI connections. The vmware boxes have different nic's as well (Broadcom and Intel), so it is not a vmware issue. I have worked with vmware to eliminate that as being the issue, and now have a call with Equallogic. The issue is everywhere, but it is more apparent with the new SAN/vmware set up.
Just a follow up after some troubleshooting - it appears the network is taking some sort of "hit" periodically. I have to narrow down where it is coming from, but it happens once or twice a day, and spikes all VM processor, network, disk, and RAM usage significantly when it happens (not constantly). I cannot say it has to do with the switches at all, but there is something happening. I have replaced almost everything over the 2 years this issue has been present.
EDIT: Looks like the "hit" is from a Symantec virus update done simultaneously to all clients. This is not the issue I am having with overall latency though.
You mention that you have those two switches stacked, and that you also have several LAGs going between them? Those LAGs are not being used, in fact spanning tree is probably taking those links down altogether. When the switches are stacked they basically act as one switch, so your current LAG setup is like plugging the switch into itself, which can cause loops. Spanning tree will prevent those loops, usually, but I could see it causing some latency issues anyway.
Try either just stacking with your stack cables, or don't stack at all and just use 1 LAG with however many links you would like between the switches.
Let me know how that goes,
Christopher Ebert - Advanced Network Support Engineer
Thank you for your reply. I have the 2 switches stacked via the 10GB stacking ports. I have LAG groups going to the devices, such as vmware boxes, uplinked switches for clients, portchannels on servers, etc. Since it functions as 1 switch, I would typically use 1 port from each physical switch to create the LAG groups to the non-switch devices for redundancy. Is this the wrong way to do it?
Introduction:Topology Diagram:Configuration Overview:Related
Information: Introduction: This document describes how to connect SG300
with Catalyst switch via STP. Spanning Tree Protocol (STP) is a Layer 2
protocol that runs on mainly on switches. The spec...
The Sx500 Series Stackable Switches offers different port features. You
can add security to a port, make the port more energy efficient, map a
VLAN to a port, make a port available or not to a specific network
portion, and so forth. The next set of articl...
On a technical level, Cisco 300 Series is far superior to the HP 1910
and 2530 products. This document provides evidence that Cisco has >100
features not supported by the HP platforms. Check this out and decide
for yourselves. Notes (2/5/2014):- Informati...