A hidden danger for your datacenter lays lurking during the holiday season – and it’s ready to bring your servers to their knees. Employees take a smartphone to family gatherings to capture the smiles of their family members. When the employees return to the office, those photos are synced and shared across your network. And, come the first working day of the new year, they’ll all be backing up at once!
Sysadmins who are just now implementing a backup solution or expanding an existing one may not be familiar with an annual phenomenon that rolls around at the end of each year. The deluge of holiday celebrations collect in a cacophony of capitalists carnage, out-of-office email etiquette, and social media madness. Unbeknownst to you (or even beknownst to you), the cameras of tens of thousands of smartphones are capturing every employee’s images of the family holiday fun, generating terabytes upon terabytes of data: Uncle Steve unwrapping his new lava lamp, babies screaming on the laps of disconcerted Santas, the three-generation picture with Grandma glowing at the children.
These phones are synced to laptops which are connected to your corporate backup infrastructure – and you may well imagine what happens next. When the first full workday of the new year rolls around, all those laptops, chock full of gigabytes of new photos, all start backing up at once across your nationally or internationally distributed network!
The cascade and then the crash
In 2015, Holiday Camrea-geddon will take place on January 5, the first Monday in January. Most of the corporate world goes on vacation, not backing up any of their data, for much of the month of December. During this time, holiday and New Years’ celebratory selfies, group photos, and snaps of happy kids will balloon the used storage of corporate and BYOD smartphones. Because the modern workforce takes their corporate laptops wherever they go, all those photos are dumped onto these computers to share with friends and family or simply to make room on their smartphones after completely filling them up.
Then, at approximately 9:00am on the first Monday in January, in your respective time zone, these photo and videos will start to back up at the same time!
While some backup systems (like Druva inSync) ordinarily take advantage of deduplication, which minimizes the server load. But with all these brand-new photos (saved at the highest resolution, naturally), this simultaneous backup across the enterprise occurs with absolutely no benefit of deduplication. Each and every block of data during this initial backup is completely unique.
Is your network ready? Is your storage ready? Is the server ready?
Better put on a big pot of coffee. It’s going to be a long day.
The On-Premises Pain
If you work in a large company with the backup solutions residing in the datacenter, you likely have experienced this phenomenon. It’s been described by some as “the worst day of the year for their infrastructure load.”
You know the painful process of trying to get the services up-and-running again after it crashes. You may find yourself struggling as the servers continue to crash as all those endpoints relentless try to shove photos of pets in Santa hats and bad Christmas sweaters through your precious network pipes and clog up the storage pools.
It doesn’t have to be this bad. Here is a simple list of things you can do, ahead of time, to ensure that your servers do not suffer the wrath of Holiday Camera-geddon. Taking these four simple steps to help your team start the new year off right.
Check your holiday file-exception shopping list. Add native mobile device image and video file formats to the exception lists of your backup profiles. By excluding these files from being backed up in the first pace, you’ll completely avoid the deluge of data come January.
Preemptively set up staggered backup schedules. By preventing users from backing up their files at the exact same time, you can reduce the network and server load that could bring down your services.
Throttle end user bandwidth. Set the slice of network resources smaller for each user during the first week back from vacation. Better to make the pipeline temporarily smaller than to endure downtime.
Reduce parallel connections. Allowing only a smaller number of users across your network to backup at the same time prevents the congestion that you’re trying to avoid.
Another selling point for the cloud
Sure, your WAN bandwidth will still be taxed if you have a cloud deployment. But the Web scale cloud infrastructure of your provider means you won’t have the same storage consumption concerns or storage server resource constraints.
To save yourself and outbound network traffic headache, and to provide your end users with the best possible experience, it’s still a good idea to implement each of the steps listed above. While not required, it sure will make everyone’s life easier if you do it.