We have made reasonable efforts to make our environment redundant and highly available. We have discussed various methods for making server workloads and storage as available as possible. However, we cannot forget about the cables and various networking gear that connects it all together.
In a well-designed network; you typically have a MDF or “core” or “center” of the network. You typically spend a little more money, and put forth and little more effort to get higher end-gear. You have lots of options:
1) Chassis-based core with redundant CPUs or supervisors
2) Multiple-chassis based core with multiple links and “failover” technology
3) Individually “stacked” switches with looping technology to increase backplane/capacity
At the end of the day, for most of you reading this, you will be performing a set of processes at the core including inter-VLAN routing, access control identifying who can go where, QOS to identify traffic, classify it, and control it. You want to make sure your core is highly available, so good choices are required.
But what about the rest of the network? Surely there is other network gear at play. Other IDF or wiring closets, or other locations around the city or country or world, right? It’s great to have a highly available core, but you also have users that need to work. There are a few ways to help keep these “edges” or “leafs” of the network highly available.
1) Multi-homed: run multiple cables
2) Let Spanning Tree figure it out. Spanning Tree is a technology to prevent switching loops. Modern networking gear is smart enough to notice a loop, and then choose a single path to get from A to B. Better yet, “tune” Spanning Tree so data flows from A to B in the most optimal way, but yet, can failover in the event a link goes down.
3) Bond multiple cables: Using Cisco Etherchannel or Standards LACP, you can aggregate your bandwidth, but if a link goes down, you still have access from A to B.
One thing we cannot forget about through all of this switching discussion is the fact we have to be able to route, too. The time will come, in about .000003 seconds, when you need access to the Internet. You want your “inside out” network to have some sort of highly available and redundant routing either between VLANs, to other locations or to the Internet. The two most common technologies are:
1) Cisco Hot Standby Routing Protocol (HSRP)
2) Standards Virtual Router Redundancy Protocol (VRRP)
Both of these basically do the same thing: two (or more) pieces of network gear “share” a common virtual gateway, which fails over near instantly, and then that gateway is where you point your various VLAN traffic. Servers get a redundant gateway, users get a redundant gateway, and the “next hop” to another location or the Internet gets a redundant gateway.
It’s extremely important for the safety of your data and your productivity that someone on your staff or payroll knows this stuff. If you’re not sure you’ve got someone on your team that can handle this, contact us today to learn how we can take this over efficiently.