Home    Bloggers    Messages    Polls    Resources
Tw  |  Fb  |  In  |  G+  |  Rss
Michael Steinhart

Enterprises Cite Legacy IT as Cloud Impediment

Michael Steinhart
Page 1 / 2   >   >>
dstrait
dstrait
8/15/2013 4:43:34 PM
User Rank
Platinum
Re: Extending life commonplace
This is true. At some point, it's not worth the effort to tidy up. (If I could teleport my messy garage into my next house, I'd probably do that (if it were cheap enough).) BUT, any sort of choke point can be useful to remove dross from systems. That can be a classic upgrade or rewrite, migration to a new data center, move from physical to virtual hardware, etc. Often, it's easier to tack on a few weeks and some budget to another project than it is to create, run and finish a project that "just" improves the system.

Another angle on this is that people don't like paying for "maintenance" work, but they will pay for "features". Therefore, if you can get some of your "maintenance" stuff done while providing features that your client sees/wants, then you will both be better off.

50%
50%
dstrait
dstrait
8/15/2013 4:37:26 PM
User Rank
Platinum
Re: There is No Substitute for Grokking Your Systems
There really isn't any substitute for "writing it down". I've yet to see any product that can automatically discover dependencies between data stores. The tool that you use to "write it down" might vary, Word, Visio, Excel, OneNote, Notepad, Google Docs or some open source variant on those tools would be OK as long as all of the stakeholders have access to those tools.

There are ways that you can minimize your risk, like publishing a set of definitive interface ("Thou shalt use these particular views and never touch the actual tables." or "Use this web service.") and then making sure that you don't break your interface when updating your programs. 

I've seen entire projects get stopped in their tracks for fear of breaking "something" out there in the userspace.

50%
50%
jagibbons
jagibbons
8/1/2013 2:34:44 PM
User Rank
Platinum
Re: Good advice
Our overall scale is much smaller than many datacenters, so what we've done may not apply to larger organizations. We are using "legacy" hardware for things like roadshow events that need equipment but aren't mission critical after the event has ended. Likewise, we have redeployed older servers with additional drives for specialty storage, like a working media share for our graphic design folks, so that doesn't eat up higher cost flash and fibre channel storage in our SAN.

50%
50%
Michael.Steinhart
Michael.Steinhart
7/31/2013 11:18:13 PM
User Rank
Editor
Re: There is No Substitute for Grokking Your Systems
It sounds like you're speaking from experience, dstrait. What are the best ways to document interdependencies? Are there particular tools that you like, or is it just a matter of writing it down and maintaining a regularly updated file?

100%
0%
Michael.Steinhart
Michael.Steinhart
7/31/2013 11:16:22 PM
User Rank
Editor
Re: Good advice
Phased migrations are always smartest -- at least, that's the conventional wisdom. In speaking with datacenter managers, it sounds as though they pick one pain point -- backup, let's say -- and then expand from there.

Have you found ways to 'inject new life' (or squeeze a few more months or years) from legacy hardware?

100%
0%
Michael.Steinhart
Michael.Steinhart
7/31/2013 11:13:54 PM
User Rank
Editor
Re: Extending life commonplace
That's a good point, dstrait. But how open to change are these operations? I've heard about apps that are a decade old but still in use, and nobody wants to mess with them. Is this all about culture, or are there rules of thumb?

100%
0%
jagibbons
jagibbons
7/12/2013 2:28:15 PM
User Rank
Platinum
Re: Good advice
is it possible to sync a cloud migration with that schedule? Or are most IT shops not that organized?

Michael, I'd guess most smaller shops are not that organized. Unless they've previously had a major infrastructure overhaul, they likely have a piecemeal approach. That can muddy up the waters on when to move. Or, it means that the legacy hardware still has life left and can really contribute.

100%
0%
dstrait
dstrait
7/12/2013 1:23:45 PM
User Rank
Platinum
There is No Substitute for Grokking Your Systems
The upstream and downstream interdependencies between systems have a tendency to increase over time and are often not well-documented because they fall into the cracks between teams, plus documentation is often an afterthought. Moving a piece of one of these subsystems from a legacy data center to a public or private cloud may cause enough latency to affect the behavior of one or more subsystems in the same way as moving from one availability zone to another.
 
The best ways that I have found to minimize this risk are to document interfaces between subsystems and keep that documenation current, ensure that you have people who have deep understanding of ("grok") how data flows between systems and make sure you understand what you are moving before you move it.
 
If you don't have good documenation, you are relying on what remains in the memories of your people. With today's outsourcing/contracting view that looks at workers as fungible entities with only short-term stints at companies or who serve a large number of clients, relying on people often has a poor outcome. Not to mention, people's memories fade. Trying to discover all of the interfaces between systems can be very difficult, particulary when there is that one report that gets run annually...and goes straight to the desk of the CEO.


100%
0%
dstrait
dstrait
7/9/2013 11:04:11 PM
User Rank
Platinum
Re: Extending life commonplace
How do you feel about moving outdated OS into a shiny new VM? I prefer to move onto fresh installs of everything, when I can talk people into it, because it provides an opportunity to filter out stuff that is no longer relevant (old logins, old data, old jobs, etc.). But almost anything is better than running on old, creaky, out-of-warrantee, ready-to-die hardware.

100%
0%
Michael.Steinhart
Michael.Steinhart
7/9/2013 2:19:45 PM
User Rank
Editor
Re: Extending life commonplace
Ouch - that sounds like a nightmare. I've mentioned in the past that I was brought in post-facto to document an application known only as "Harvey's Tool" when Harvey was transferring overseas. It took about 2 months, and only then did IT begin the process of migrating it from Harvey's shadow VM back in-house. And this was back in 2006! I was hoping things had improved since then.

So how long do these legacy systems stay in place because of inertia? What's the oldest/most outdated legacy system you've ever had to deal with?

100%
0%
Page 1 / 2   >   >>
More Blogs from Michael Steinhart
After a 10-month run as editor in chief, Michael Steinhart moves from The Enterprise Cloud Site to another UBM publication.
The competing giants aim their products at different segments of the enterprise IT world.
According to our last poll, most of our audience is happy to attend IT expos and trade shows, as long as the registration fees are waived.
SolarWinds head geek Lawrence Garvin explains why companies must increase IT security awareness across all employees and departments.
A new report shows that IT professionals aren't happy with their security information event management (SIEM) tools.
Digital Audio
Latest Archived Broadcast

Security expert and author Ira Winkler will be the special guest of The Enterprise Cloud Site Radio Show, discussing cloud security, on Thursday, June 26 at 2 pm (EDT).

flash poll
Live Video
On-demand Video with Chat
What lessons can today's enterprises learn from federal cloud intiatives -- successful and less so?
The Enterprise Cloud Site
follow us on twitter
The Enterprise Cloud Site
like us on facebook
The Enterprise Cloud Site
About Us     Contact Us     Help     Register     Twitter     Facebook     RSS