Thursday, October 25, 2012

-- Basic knowledge about some Network Artifacts --

As it had happened to any other DBAs out there, our roles have been evolving and mine hasn't been the exception. I recently have been more involved with Network architectures, I am sharing this information with you hoping that it can be useful.

VIP (Virtual IP): It is an IP that is not bound to a single physical interface.

LB (Load Balancer): Artifact that allows you to share network requests (following some algorithm) across multiple resources.

GSLB (Global Server Load Balancing): Allows you to route network traffic to multiple locations. It's main objective is to route your connection to the most efficient server available to you. 

GSS (Global Site Selector): This is a Cisco GSLB Product.

GTM (Global Traffic Manager): This is a F5 GSLB Product.

Broadcast: When you need to have more than two nodes see the traffic on the same subnet.

Unicast: Unicast is used when two network nodes need to talk to each other.

Anycast: With anycast you advertise the same network in multiple spots of the Internet, and rely on shortest-path calculations to funnel clients to your multiple locations. As far the network nodes themselves are concerned, they're using a unicast connection to talk to your nodes

Multicast: Is like a broadcast that can cross subnets, but unlike broadcast does not touch all nodes. Nodes have to subscribe to a multicast group to receive information.

| TYPE      | ASSOCIATIONS     | SCOPE           | EXAMPLE |
| Unicast   | 1 to 1           | Whole network   | HTTP    | 
| Broadcast | 1 to Many        | Subnet          | ARP     |
| Multicast | One/Many to Many | Defined horizon | SLP     |
| Anycast   | Many to Few      | Whole network   | 6to4    |

Developers vs. DBAs who will win the battle?

In the last 9 months I have been involved in several battles with the development group, and before I continue let me be clear DBAs are humans and we do mistakes. I have tried several approaches to those issues, but only 1 has provided me the satisfaction and the benefits expected.

"A natural approach is to say, I didn't do anything, the problem is not in the database, it is in your code, and it is not my business since I am not a developer. "

As a DBA we tent to think this way. "I did not change anything in the database, I see no errors in the alert.log therefore is not my problem. "

My question is, for how long are we going to say the same, over and over, even that is technically true. I lose a tremendous amount of time in useless RCA meetings (Root Cause Analysis) where everybody points the finger to everybody else, and people worry more about who to blame it to, instead of how are we going to get it fix for good and avoid future recurrences.

Let me be clear, "No one in a organization, DBA nor Developers wants for outages or problems to happen", if there were no problems, we will have no jobs, therefore problems will happen, Focus on learning from the problems and avoid them in the future.

The approach that has worked for me, and have provided the satisfaction and the benefits expected is to research about where is the problem, even that it may not database related. Yes don't be afraid to put your nose into somebody's else business (I know it does not sound right but you know what I mean with it ;-) ), you are getting affected so they need to understand clearly your interest.

In the past I have found that some of the problems are:

- The developers are been push by the project managers into meeting unrealistic and
unnecessary deadlines, and they sacrifice quality for meeting the deadline, in other words
if it works is good enough, I don't care the resources it will consume or how long will take to

- Very inefficient spaghetti code produced for a constant change in the scope of the project.

- Lack of coding standards, and in this case the DBA produced coding standards for them and
the DBA is the watcher that those standards are met. (Yes, I know it is not in our job
description, but it works).

- The developer does not know how to code, they just know a reduced set of instructions and
try to program with whatever they know, therefore producing tons on inefficient code. For
this case we are providing the developers with a database architecture workshop and a SQL
and SQL tuning workshop. In other words I have to help them so they can help me. (Yes,
another one that is not in your job description).

Conclusion: Don't be afraid to think out of the box, rather than complaining about the problems
try to find out where are the problems (even if they are outside your
boundaries) and try to be part of the solution.