Backup Reporting Myths

Back in June at Symantec Vision we had a group that busts myths perform (we're not allowed to use their official name but you can guess who these "busters" were).  We love to throw claims (myths) around in the storage industry so I thought I'd debunk a few things I've been reading about lately.  Yes, I am writing from a storage vendor as well so take it with a grain of salt but this is an attempt to at least bring these claims to light and let you make your own decisions.

Real time Reporting
There are several third party products that claim to do this.  These products typically compare the most with our own Veritas Backup Reporter (VBR for short) but really this is a feature we compare with NetBackup Operations Manager (NOM for short).  Both the 3rd party products as well as VBR simply can not do true real time reporting.  The way backup data is collected from all these applications is by executing a NetBackup command line: bpdbjobs, bpmedialist, vmoprcmd or reading logs information.  The command lines spit out a bunch of data that is parsed and then stored in a database or the log messages are read in and stored in a similar way.  These command lines and logs are looked at every few minutes or hours as to make sure new data is captured as things change in NetBackup.  At no time is the backup application giving the information to the reporting application proactively though, the reporting application always has to ask first.  That is why it is not real time.  You can try to run the command lines more frequently to simulate real time but that can potentially have a performance impact on NetBackup - something we've seen a number of times with the 3rd party tools at large NBU installations.  For this reason we do not recommend configuring VBR to collect more frequently than what it defaults to which is different for the various types of data but not typically more often then every 15 minutes.  That's what the NetBackup engineering team has recommended and we're sticking with it.

Now there is one product that is actually real time and that is NetBackup Operations Manager (NOM).  NOM, if you're not familiar with it, is an add-on that comes with NetBackup.  If you haven't tried it yet, give it a shot, you've got nothing to lose - it's FREE.  The NOM engineering team has been adding features and scalability to it since NetBackup 6.0 when it first was released and now at 6.5.x we're starting to see much greater adoption of it.  NOM does not collect data through command lines, it has a special interface in to NetBackup that is event driven.  In other words when something happens (drive goes down, job finished, policy changes) NetBackup proactively tells NOM about it, NOM does not have to ask.  This is a functionality that can only be delivered by the backup application itself and not a 3rd party.  So get NOM.  It's cool.  Trust me.

Web 2.0
I could go on a whole rant about the use of the term Web 2.0 in general.  Frankly if you're developing a modern web and not using AJAX, W3C standards, widgets and things of that short you should have your pulse checked.  Both VBR and NOM have a core of HTML/Javascript that drives them.  We've always stayed away from Active X applets, Java applets or anything else that is more heavy weight.  The latest release of VBR caches and loads all its reports and dashboards through AJAX so don't believe that no one else is doing Web 2.0.  We all are since it is the defacto way to develop a web application now a days.

Predictive Analysis
Another one of VBR's competitors loves using this term.  It sounds fancy but similar to the Web 2.0 term all the backup reporting applications have this.  We just happen to call it "forecasting" or "trending analysis" and you can do it with VBR on every single report that there is.  One of the nice things about a backup reporting application is that you get a lot of data eventually.  And with that data you can start to apply mathematical analysis to see where you're going to be in the future.  It is probably one of the biggest single reasons people look at getting a backup reporting application in the first place.  Now another detail when people use the term "predictive analysis" is that it is usually backed by a statement about how information is collected from lots of pieces of the environment and then put together to determine what the root cause of the failure is.  Sounds awesome.  Kind of.  Applications that collect information from lots of different devices sound good at first but when you start to look at an HCL for them it is a fricken' nightmare.  You have to make sure the firmware is correct for each one (will the storage guys lets you back rev the array firmware just for reporting? ever try back reving tape drive firmware?) and upgrades often break or change the way data is collected so you are in an endless cycle of patching or upgrading either the reporting application or the devices it collects information from.  I'm a pretty firm believer that this is one of the fundamental reasons why a true heterogeneous reporting (SRM type) application never took off.  But even if there was enough engineering time and compatibility to do this at the end of the day what does it really get you?  The fan speed and temperature of your tape drives?  Who cares? You probably want to know about failures or status and guess what - the reporting application knows about that already anyways.  So with VBR we've taken a kind of philosophical approach where we go deeper to collect information through NetBackup instead of going wide and trying to keep up with all the devices that ultimately don't give you any more information than you can get through the backup application.  It's depth versus breadth.

So I encourage you to take a look at backup reporting and obviously being with Symantec I'm a big advocate of Veritas Backup Reporter.  To find out the details sign up for a 1-hour demo webcast at symantecevents.webex.com and we'll make sure we don't use too much marketing lingo and buzz words.