|The standard enterprise data center, not.
The Eastern tip of Long Island is home to one of the United States’ largest wine producing regions. For anyone who has explored the wine trail and the abundance of wonderful vineyards, you may have also noticed something equally wondrous to see. Once you pass the town of Riverhead you encounter huge fields of bright green perfectly manicured grass. Alas, this area is also known as one of the largest producers of sod for the Northeast, with some fields stretching for miles uninterrupted.
If you’ve seen how sod is farmed, it’s quite amazing, as the manufacturers literally roll up the grass, roots and dirt like a rug.
For infrastructure architects, there is nothing quite as exciting as designing and building pristine networks and compute farms. At Apstra we have customers who regularly deploy new data centers and get to experience the value of the Apstra Operating System (AOS) Intent Based Network System, but we also have potential customers who want to try AOS but don’t know where to begin.
AOS has a number of features that can be used for operating a legacy “brownfield”, but the amount of value and simplicity offered when deploying AOS on a new environment is truly something to be experienced. When an architect or operator uses AOS to design, deploy and operate a network, they see the new paradigm for system management and quickly realize how inefficient the old methods are.
So in a modern enterprise IT shop, how can an architect dip their toes into Intent Based Networks?
Playing Infrastructure Leapfrog
All data centers go through a lifecycle, where they rip out old equipment and replace it with newer, more powerful gear. I like to remind customers that if their servers are designed to be replaced every five years, then one sixth of their data center should be upgraded or replaced every year. I describe this as Server Leapfrog. In reality this process is much more complicated, it can take a whole year to evacuate services from a compute pool. If you have 6 rows of compute, you end up with four rows with no changes for the year, one row that is being decommissioned, and one row that is being rebuilt with new hardware. This formula is consistent based on the hardware refresh cycle, For example, if you operate six fields of rows, the model is generally the same as if you only have six racks.
|Once Rack1 has been completely decommissioned, Rack6 can be built in the same location
Rolling Out Your Green Patch
|If only networking and server deployments could be this easy
A green patch is a new system or “pod” that is a fresh build with some freedom to try new hardware, software, and system designs, A green patch can be as small as a single rack. With AOS, a green patch can be two switches (one leaf and one spine), however for the purposes of this discussion we will focus on a
2 by 2 Clos Leaf/Spine design to ensure adequate redundancy.
Using two top of rack (TOR) switches and two centralized spines, you can achieve a much lower price per port than a large chassis. Buying a chassis requires a huge upfront investment, as well as logistical issues related to the size and power requirements of a large switch.
Your new green patch environment with four switches only requires four 19″ rack units (RUs), so it will easily fit into the top of one standard rack with plenty of room for servers. These fixed form factor switches also use standard 110V power, so they can be placed just about anywhere in the data center.
Despite the fact that we only have 4 switches, AOS allows the architect or operator to easily expand and contract the network based on need. If you need more switch ports for servers you add more leaf switches, if you need more bandwidth between servers you add another spine. AOS automates the configuration and deployment down to telling you exactly how to plug it into the rest of the network to eliminate the possibility of service interruptions. You can also choose from a large variety of different vendors for the hardware, allowing you to add new capacity at the lowest price point and mix and match different vendors.
Intent Based Networking solutions wizards, you can pick from a large number of common vendors and switch models, dial in the exact subscription ratios you need, and AOS will immediately provide an accurate cabling list. I like to think of this as the instruction manual that comes in every box of Legos. Once the system is connected, AOS will automatically validate all of the cabling to ensure the system is wired properly, and then you can deploy the running configurations to the devices. AOS will automatically monitor and report on any issue it encounters in the IP fabric, as well as inform you of any changes it notices in the configurations.
Using AOS to design, deploy and operate your new green patch, you automatically gain the following capabilities:
- A single dashboard for the health of the system
- A simple GUI for managing switch ports, servers, vlans, routing, and more
- Configuration validation and deployment for all devices
- Automated monitoring and alerting of all major issues that impact service availability and violate your stated intent.
- Easy point and click scaling when adding or removing new leafs or spines
- Integration via REST API to all 3rd party systems and tools
I hope this brief explanation has given you an idea of how quickly you can build a new network with AOS, and how small a running system can be. There are a number of additional features and value that AOS provides and we would love to give you an overview of the benefits that come right out of the box.
But rest assured, you do not need a lot of resources to build a brand new pristine network. It can indeed be as simple as rolling out a small carpet of bright green grass.
There’s nothing like fresh green grass beneath your toes. Reach out to us to help you build a new special place in your data center.