Re: Oracle DBA to PostGreSQL DBA?

From: Tim Gorman <>
Date: Sat, 17 Dec 2016 11:15:30 -0700
Message-ID: <>

No, there *is* indeed something to see there, as it is an example of an IaaS migration. Here is a question to ponder: What is the impact of this migration on the SysAdmin team?

IaaS migrations are transitional, for both parties (i.e. cloud provider, cloud customer). Cloud providers are scaling up and improving their services, transitioning from leading-edge to mainstream and enterprise. Cloud customers are testing, often committing.

I think the present and near future holds a lot of transitional "hybrid data centers", with production remaining on-premises until the current leases expire or the current purchases amortize, and development/testing and other non-production systems leading the charge to IaaS. In a few years, after non-production systems secure the beach-head in IaaS, production systems might follow the established migration path, or they might evolve directly into SaaS migrations by then.

If past history is any guide, consider these rough facts...

The first wave of mainframe-based commercial deployments covered about 25-30 years, from the 1960s to the 1980s. After the advent of desktops and departmental computing in the 1980s, client-server computing gained momentum in the 1990s, over a period of period of perhaps 8-10 years.

Web-based 3-tier architectures came to dominate on-premise in the 2000s, and then co-location and/or hosting of these architectures arose in the 2010 timeframe. The problem with co-location/hosting services is that it is continues just-in-time, budgeted provisioning of compute and storage (i.e. slow).

In the late 2000s, AWS was experimental, but in the past 5 years, IaaS services have begun to dominate. The main difference between co-location/hosting and IaaS is instant provisioning and pay-as-you-go pricing (i.e. agility).

So ask yourself: What is next along this continuum? When? And where am I positioned?

More practically: When do current hardware leases expire? Or amortize?

 From a personal standpoint, after a really painful experience with a startup going out of business in 1988-89, I made four career resolutions by 1990...

  1. never work where technology isn't the product
  2. nothing good happens after 5pm
  3. data is the only thing with value
  4. pursue a technical, not managerial, career path

The first meant that I would never again be a "cost" to the business, I would "be" the business. The second meant that I wouldn't work for poorly-run or "evil" companies; consistent demands for long workdays are a leading indicator of company failure. The third led me away from application development and toward databases. The fourth has kept me happier, if less upwardly mobile.

Like the US constitution, I've made amendments along the way, but so far it has been useful.

On 12/17/16 02:24, Norman Dunbar wrote:
> For what it's worth, my current contract is a migration from Solaris
> to Microsoft Azure cloud.
> Once I'm done with that, the permanent DBAs are staying on and taking
> care of it all. It's "just" a server really. In this case anyway, but
> a server you can "turn up to 11" for some stuff and back down again
> for other stuff.
> Nothing to see here!
> Cheers,
> Norm.
> --
> Sent from my Android device with K-9 Mail. Please excuse my brevity.

Received on Sat Dec 17 2016 - 19:15:30 CET

Original text of this message