The VA Trip (2022)

Zesty
Zesty
Published in
3 min readAug 31, 2022

As Zesty users know, Panduhz and I (David) recently went to Virginia to deploy backups, RAM upgrades, CPU upgrades and network upgrades. What users don’t know is that we did one months' worth of work in just 15 days. To complete our upgrades in time, Panduhz and I worked 10-to-18-hour days. After a few days of sleep deprivation, malnutrition and just pure work had kicked our butts but we had to keep pushing knowing our time was limited.

This was a scheduled trip since April from August 1st to 31st. Due to family and friends, we had to cut the work trip short to August 19th. Something unplanned for was that August 1st and 2nd were big drop days, we couldn’t do upgrades due to drops happening. This is the case where we can’t touch anything server wise as clients need to be online 100% of the time. Those two days we spent helping clients in tickets and monitoring network as usual. August 18th, we spent doing some final touches and we were out.

Upgrades

At first the plan was to just do our RAM upgrades and deploy backups but then we figured we should do CPU and network upgrades also. This was our mistake as the workload turned out to be too much.

RAM upgrades to give clients 50% more RAM.

Good news is we did make it through all of the RAM upgrades by the skin of our teeth. Now, it was onto network upgrades and storage node deployments. Network upgrades are not as temperamental as RAM upgrades because it’s simply just racking switches and plugging in more cables to the NICs then just configuring the switches and servers into a LAG configuration. This will allow us to distribute loads across different backbone switches and setup our nodes for 80G worth of capacity instead of just 40G. Clients will be able to burst higher than ever with this setup while being able to feel secure due to us having a second network deployment per server.

Arista switches for backup/LAG configuration.

Storage/Backup Deployment

Storage node/backup deployment will allow us to retain all client data for 48 hours. All backups will be run during off hours, so we don’t congest any disks/network capacity. Client backups will allow us to run all nodes in a RAID-0 configuration for our Samsung NVMe drives. This will give clients a small performance boost from 4000MB/s read/write to over 6000MB/s read/write.

It may not look like much, but this is over 2.5 petabytes worth of storage.

Image Node Deployment

Image nodes will be used for Windows Server 2012R2/2016/2019 deployments for all VMs going forward. This will also allow us to create custom templates with pre-installed software to help you with easy deployments and setups.

Food

For those of you wondering yes, we treated ourselves out often but due to long work hours we were only able to fit in a decent meal a day.

Treating ourselves out a bit. 😛

Final Remarks

Even though we didn’t think we’d meet the timeline for all of the upgrades, I am happy to say that Zesty was able to upgrade over 300 machines worth of RAM, deployed network backups + extra capacity, and setup storage nodes for complete redundancy of our servers. Panduhz and I pushed over 190 hours of work in just 15 days of work.

--

--