Recommendation for future puredisk integration/versions/patches

[alazanowski]

First off i wanted to complement on the efficient deduplication that puredisk offers- when it is working without any bugs, i am very pleased with what its capable of doing.

However, there is some recommendations i need to make in regards to future versions that is coming from an enterprise view of the environment.


Recommendation #1 - I am extremely frustrated with the fact that server agents are based uniquely off of an agent ID. I understand that its following best practices for SQL and that there is a "security" implication on basing things off of the name of the machine, but it is EXTREMELY frustrating in an environment that revolves around thousands of clients to manage that are constantly being reimaged/rebuilt and redeployed. While i understand that there is a command line on the silent portion to relink an agent with a particular id on re-install, if i have to do this on 10 machines a day and have to modify my deployment task for each individually, that is not a very practical method to handle. If its something where you cannot allow that to happen based off of their FQDN, could we at least gain a utility or portion for the GUI that would allow us to "merge" agents into 1 id? So that if i have store1.domain.com (agent id 2) that got rebuilt from a new image, and then puredisk installed now with new agent id 3, then i could take store1.domain.com (agent id 3) and store1.domain.com (agent id 2) and merge them into one agent and default to one id.  I spend too much time a day providing these administrative functions and making it more difficult to keep it up is not making its practicality show.

Recommendation #2 - PLEASE provide an alternative to the web gui. I know you have a new command line interface in 6.6 but its not as easy to navigate. My preference would be for a java based gui we install locally to run off so that it runs much smoother and more efficent, like the netbackup java gui interface. I know this idea of cloud computing and web interfaces is attractive to some because it means not installing anything, but if i had to spent 5GB to install something locally for better performance i would - and i know plenty of people who would as well. The new web gui is much better than the old 6.5 but it is still not as high in performance. I would also think maintaining a web gui is much more difficult to test across multiple browsers and versions - for example, in internet explorer i cannot use half the functions because it'll open its seperate window and ask for me to authenticate a certificate but unfortuantely wouldnt allow me to expand it to click it (hard based window). So i began to have to use firefox instead, but then i see issues often where the drop down menu wont display correctly. I've yet to have an issue with the netbackup java gui that i've installed, even with different versions of windows.

Recommendation #3 and maybe more of a query ---- With the integration with netbackup 7- will the client side deduplication & backup present itself as 1 backup image to the puredisk metadatabase? One of the issues we could potentially see is significantly limiting our versions based on the limit from the metadatabase. We have folders that have literally thousands upon thousands of very tiny files that need to be backed up, and if you span that across hundreds of stores, that would significantly limit the amount of versioning you could have. If that wont be the case, maybe design the PDDO option in the policy to do only 1 full, then incrementals only after that, with synthetic fulls occuring afterwards.

More recommendations to come as i think of them, but i think the potential for puredisk could be huge with a firm integration with netbackup and resolving of some common problems.

Thanks!