Updated: May 28
BMW or Roller Skates today?
Backup and recovery from disasters is an interesting topic.
It is in fact rather fascinating on many levels that often have me scratching my head in wonder, amazement and confusion - often in the same customer conversation too!
I have had a career long relationship with backup systems of many flavors, starting with IBM T series tape libraries for Mainframes to current day solutions and schemas from Data Domain, Rubrik, Cohesity, Avamar, Networker, Tivoli, Commvault, Netbackup, ARC, HPE, StorageTek and other players like Barracuda or Dell dense storage targets, to name but a few.
I even dabbled with IBM guys on storing vast amounts of data on crystals of various kinds at one point in my backup and restore adventure.
Newer players like Veeam and HYCU offer interesting platforms to consider in this BC/DR mix these days as well.
My most educated BC/DR customer once got rid of all of their Tape libraries and replaced them with several clusters each consisting of 4 Hitachi Data Systems VSP series storage arrays - all running RAID 10!!
This was just for backup!! I kid you not!!!
I stayed for a week just to marvel and verify that this madness was actual reality.
It did seem like an acid trip at first glance....
Fascinating is what that was!
This drastic engineering solution was prompted by them losing several legal cases because they could not produce evidence and data from backups within the window the court gave them to produce said artifacts in question.
On the other end of the backup solution spectrum I have worked with other Healthcare organizations for whom backup was and still is just a tick in the box to show they made some due diligence effort at long term archive of data but ultimately they did not really take it that seriously.
Most of my customers fall between these two extremes with the RAID 10 on HDS VSP clusters being one rare but exotic solution to the problem versus Tape archive stuff carted off to Iron mountain by the truck load for the ones on the other side of the backup spectrum.
Those backing up to tape know full well that they cannot recover from tape reliably but check the due diligence box along the way regardless by default.
To me, how customers back up data tells me all I need to know about that organizations IT strategies and attitude as it pertains to data value and importance to that organization when examining their storage challenges.
It is thus my opinion, after some 35 odd years experience of deploying all manner of exotic backup systems that range from ultra elite extreme backups of immutable state type to tape backups is that users need to be presented with as many options as you need to cover said backup spectrum that offers them the choices and these choices will have reciprocal effects in their environment to suit the value they elect to place on their data.
For a Nutanix cluster the backup GUI should ideally offer RF1, RF2 and RF3 targets as Data Protection choices so that said customer can decide for themselves what they are willing to spend on their backup endeavor.
Let me stress that RF1 will never be an option for a normal Nutanix HCI cluster, not ever, only for dedicated Nutanix Mine appliances or dedicated Nutanix Objects Appliances.
There was a time when EMC had RAID 0 options on storage arrays which I must admit did disturb my tiny little synapse a whole bunch!!
Who would buy an expensive storage array and use a RAID 0 schema on it? I still shake my head in wonder over those few customers who did in fact do this.
I guess some people will buy a toilet bowl to drink champagne out of....Just because they can!?
RF1 would only ever be an option for clusters specifically doing dedicated Nutanix Mine Appliance or Nutanix Objects archive duties.
I am pretty sure a good 63% of these customers would in fact elect RF1 due to the cost implications of RF2 and RF3 and the size of storage infrastructure they would need to accommodate said desires....
So what if data corruption or a failed disk means I will lose all the data in the backup set? I chose to take that risk fully aware of the potential risks involved is what I am hearing most retort for this unfortunate posited possibility.
"To each his own" I hear from my sole surviving synapse...
This information tells you how much they value the backup data in the first place.
On the other side of the coin there is the customer who may decide the backup is so valuable they will run it all using RF3.
For those of you that do not know, Nutanix RF3 is like double RAID 10 for specifically paranoid customers who can never lose data no matter what. Not ever. Under any circumstance.
Of course this will have a Domino impact on the storage quantity and the cost of the solution you will need to plan for vis a vis such storage backup ambitions.
Although you may want to back everything up to RF3 Storage targets, your CFO and CEO may have other ideas of what the value of the backed up data is to the company wishing to deploy said backup schemas along with suitable infrastructure to accommodate their ambitions - all at matching price tags of course.
Who will assume the risk at what price point comes into play here. This is an executive level decision.
The easiest thing from a Nutanix perspective would be to offer RF1, RF2 and RF3 schemas and let the customer decide what each level will cost them to backup, with the risks clearly articulated for each choice that they happen to make.
It would be nice if the GUI showed what storage size they would need based on their selections....just sayin....
Data Dedupe and compression offer up their gains and the choices combine to reflect what size of storage system they will require with all the Data Protection, Data Dedupe and compression reductions baked into the equation and fully accounted for in what is a very complex set of equations added up together.
For a Nutanix customer using HYCU with Nutanix Objects the setup screens should offer RF1, RF2 and RF3 choices with each selection clearly showing what storage quantity is required for the backup schema chosen and choices selected with the risks clearly explained in the technical review phase by your ever caring and attentive Nutanix and HYCU SME on the subject.
I must mention at this juncture, that in my head, I MUST have the exact same car Lewis Hamilton races in each F1 Grand Prix every other Sunday.
In pragmatic reality my garage has two Jaguar EV400's and one Chevrolet BOLT EV in it.
I do not have the $20 million budget to be driving a F1 car and I doubt I could get a DMV license to drive it on the road anyways although either Larry E from Oracle or Scott M from Sun once managed this feat so there is precedence.
Plan B is a Lucid Air Dream Edition with all the trimmings for a cool $198K US a pop.
It was either two Jag EV's or one Lucid Air, so two Jags it was. One HSE and one S.
BC/DR works the same way. You buy what makes the most pragmatic sense for your budget.
Decide what gives you the biggest bang for your buck so to speak.
If badges and names for vehicle marques or luxury was not a factor I would have five Chevrolet Bolts serving my needs but we decided we wanted some luxury and prestige in the EV experience and paid for the features offered.
You get what you pay for with both BC/DR and EV's.
This is why our most recent EV addition was a white EV400 I Pace S and not another HSE.
In fact if I had been doing the buying it would have been two I Pace S platforms from the outset because my wife does not use any of the HSE features anyway.
I use them, but only because I know they are there.
Do I need them? Absolutely not!!
However, there is no point to buying or building a rocket ship if you ain't going to outer space now is there?
Anyways you would be surprised what customers do with backup strategies and decisions.
The best scenario is to offer them all and let the customer decide which boxes to tick carefully explaining the what you effectively get pieces for your choices along the way.
It can be no other way!!