Agentless Backup is Not a Myth
Tony Lock
Programme Director, Freeform Dynamics
It is correct that many new server and, especially, server virtualisation systems offer the ability to manage systems far more dynamically than ever before. Indeed, it is fair to say that until now the primary mandate adopted by most IT organisations has been "if it is not broken, leave it alone". The capabilities inherent in many systems to move workloads around very quickly or to create and take down virtual servers and their associated workloads in a matter of minutes open the door to new operational modes.
At the heart of the matter is the question of server resource allocation and how to obtain "optimal" benefits from the systems at hand. From a skills perspective there is still the need to be able to monitor and manage the complex systems at all levels of their operations. But the very idea of being able to alter the configurations of systems as a routine part of standard operations will eventually demand that new skills, or at the least operational procedures to be put in place.
Chief amongst these will be some way of selecting what proportion of available resources is allocated to each application in the dynamic infrastructure. Clearly if the infrastructure has more resources than can be consumed under any likely workload scenario, there will not be too many demand clashes to handle.
But if - as is likely - organisations seek to limit the overall size, and hence cost, of the IT infrastructure, there will be times when someone will need to decide how constrained resources are allocated. This will require good "vision" of the likely business consequences of such choices.
Thus the IT department will need to possess the skills and monitoring tools to make such judgments, or put some form of automated policy prioritisation processes in place. Getting hold of either of these will require no small measure of communication and diplomatic skills as well as a good handle on business reporting systems.
COMMENTS
@Sir Runcible Spoon
Oh, no denying there's still (currently) a market for us old-timers. Jack of all trades admins are still hugely relevant in the SME space.
There will always be a limited need for jack of all trades types...but I fear the number of us that are required is shrinking. I am sure it’s cyclical…all IT job patterns have seemed so far to be, however what seems to be happening is a higher demand for narrowly-focused, (and most importantly cheap) specialists.
Smaller organisations are more and more outsourcing their IT. Those outsourcing companies, be they Dell, HP, or your local consultancy are slowly evolving along the same path. "Customer support staff" at the edges calming the customer down and finding out the problem, passing the information back to their actual "doers" who then push the button to fix the issue remotely. In the case of a dead bit, the backup is activated, and the customer sends it back to the outsource company for replacement. (Commodity hardware has gotten to the point that this is easily doable.) It is all very process-managed and sterile.
I don’t claim that this is the best way to do things…but simply that this is the direction that our industry is heading. IT is largely no longer about innovating or solving a problem with a unique solution. Almost everything you could want comes as a pre-packaged appliance now, or can be solved with easy-to-manage nifty virtualisation tricks.
Like the handyman of the physical world, the IT generalist will probably be the only of the "IT trades" that a company keeps on staff. The handyman can fix your doorknob, change the lightbulb or coax that creaky boiler back to life. If he needs something done outside his particular expertise, he calls up a plumber to have new pipe run, or an electrician to wire a room to code. We’ll be calling up a cable monkey to run lines, a datacenter specialist to install rack/power/cooling or a programmer to add functionality to a website.
Pay attention to how many handymen even large companies employ…and what those fellows get paid. They are what IT generalists are set to become. (On a totally selfish note however, it gives me no end of glee to picture several of the "specialists" I know as the digital plumbers of the future. Sadly, the "IT Specialist" types in my neck of the woods really look down on us generalists quite a bit.)
Good luck to us all!
ssh, bash perl?
If your seriously using ssh bash and perl alone to manage large numbers of systems, even smallish numbers(more than 50) then you really should look at better tools like cfengine or puppet.
ssh bash and perl alone are far too limited to handle real distributed admin work, unless of course your writing something along the lines of cfengine or puppet in perl ..
Just last week as an example my company built a new edge location in London(I'm in Seattle). We sent one guy out with 1 server running ESX and a VM with enough data to seed the rest of the network. He configured all of the remote management cards, the network guy got the VPN up, and I remotely installed ~55 physical and virtual machines in roughly 6.5 hours over the WAN including all application configuration and everything. And that was using real remote installs, not stupid image-based crap.
It could of been done faster though I decided to throw a wrench into things and revamp my kickstart setup at the same time which made me have to go troubleshoot some things and re-install some systems to fix them.
But all in all it was a pretty good day.
Wouldn't of been possible in that time line to do it without cfengine though(I haven't used puppet hear it's pretty good). My cfengine configuration is about 20,000 lines, to give an idea of the number of variations and tests that the system runs to ensure compliance, and that compliance is re-checked every single hour of the day.
I suppose I consider myself a jack of all trades guy whether it's systems, storage, networking, automation, monitoring, and architecture. It's really rare that I physically touch hardware these days.
I don't personally know anyone else that has my skill sets, though they weren't easy to come across, made a lot of sacrifices over the past decade(well worth it in the long run I believe, no regrets).
Pick the right products and you can scale pretty damn well.
lots of parallels
My old school physics teacher complains that he cannot understand his new Honda - even though he taught auto engineering.
Same with the data center, how on earth do you know what is happening with hundreds of apps interacting on top of an dynamic, ever changing infrastructure. Talk about a house built on sand. Virtualisation/elastic environment = sharp knife = blood on floor.
Complexity has forced the auto manufacturers produce analysis boxes to pinpoint problems in their vehicles. I need one for my e2e processes but unfortunately, we are not there yet and there is little on the horizon.
Come back squawk box, all is forgiven.
IT infrastructure monitoring strategies
Agentless Backup is Not a Myth
Top 10 SIEM implementer’s checklist
Steps to Take Before Choosing a Business Continuity Partner
Enabling efficient data center monitoring
