Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
And before you begin to think that I'm trying to talk you into a truck or sedan, recognize that if you delete a file off a production server, a replication engine will very expediently delete the copy of that file off the target server in 'real time'. That point alone brings us to why tape and disk should never be mutually exclusive.
If you need to see what your files looked like yesterday, or last week, or last month, or last year, use tape. Tape is an inexpensive method of archiving snapshots of whole files over long periods, with the anticipation of restoring periodically.
If you are concerned about ensuring a near zero loss of data or productivity, you need to look past back up and restoring. Data protection is replication.
Consider if you back up your data at midnight on Monday and then have an outage at 4 p.m. Tuesday afternoon.
In business continuity terms, the RTO (Recovery Time Objective) of "when will I be running" is the amount of time that the tape(s) need to restore (plus the amount of time to get the tape set back from your offsite provider). Best case RTO is a day (i.e. so you receive the tapes on Wednesday and perhaps it is restored by end of day, but more likely by Thursday morning). With replication technology, the second copy of data is already accessible (on another remote platform) so RTO is close to zero.
Still using business continuity metrics, the RPO (Recovery Point Objective) of "where will my data be when it is available" is the latency between when the tape backed up and when the outage occurred. So, at the end of your day of getting the tapes and doing the restore, your data looks like it did Monday night at midnight. Your users, who did not have access to any data on Wednesday, will discover on Thursday morning that the data is from Monday night with everything Tuesday being lost. With real-time replication technology, the alternate server has a copy of the data that is seconds old.
I once heard someone define Insanity as "repeating the same process over and over, but expecting different results". Tape is inherently whole-file, scheduled, non-WAN-friendly and tends to require a backup window. If your only approach to disaster recovery is using tape and couriers (the repeated process), then do not expect to get different results (significant RPO and at least day-long RTO's).
All that being said, replication may not be for everyone, but then again, neither is any other tool, brand or technology in our workspace.
So, stop debating and embrace each methodology for what they do best (tapes for inexpensive snapshots over long periods) and disk for up to the minute protection of the data). And if you back up the data at the replicated facility, that is an 'off-site backup' without paying for courier services. This leads me down a path of ROI and monthly operating costs compared with a capitalized system purchase (but that is a topic for another article - maybe next month).Another path we could talk about (in another article) is discussing disk-snapshot technologies. In those cases, the storage solution captures portions of the file-set that have changed over the past X hours. As that technology continues to develop, some clients will eventually consider baselines and disk-images to be a replacement for tape and an even better partner for replication (but again, that is for another article, so check back on this).
In closing, when you are ready to take a long look at really protecting your data, consider enhancing your backup solution with replicated data (which may offer the hybrid approach that you need for protecting the productivity of your users and ensuring long term data survival).
And when you next go car shopping, consider an SUV (which may offer the hybrid approach of truck and luxury sedan that meets your needs). But not the 4x4 versions. Your friends know that you aren't going to take a Lexus or BMW SUV into the muddy hills, and will laugh at you.
About the Author
Jason Buffington has been working in the networking industry since 1989, with a majority of that time being focused on data protection. He is a Business Continuity Planner and an MCSE-MCT. He currently serves as the Director of Technical Marketing for NSI Software, enabling High Availability and Disaster Recovery solutions.
He can be reached at email@example.com