Efficiency in Data Movement Matters

on

My grandfather was a plumber, and my dad started his career as an electrician. Both men taught me valuable lessons that I use in my career today. Many of those lessons have more to do with ethics than technology, but both professions have significant commonality with data movement.

At its core, plumbing is about moving water from point A to point B in a controlled and predictable manner. If you turn your faucet on, you should expect a consistent water flow each time, with relatively negligible variations for the most part. Electrical work is functionally no different, though instead of water molecules, it’s only moving electrons. In a similar manner to the faucet, if you flip a light switch, the power should turn on or off in a consistent, predictable manner. A key to both systems working well is to make them as straight a line as you can from point A to point B.

Bends in water pipes, while necessary, should be minimized in order to ensure as smooth a flow as possible – both in the process of ingress (water) and egress (sewage). The farther you get from your source, the more work it is to retain the same pressure. In addition, for hot water flows from a traditional tank system, longer runs are left with a long lead of cooling water, wasting gallons of water every time hot water is requested and the sink needs to run to purge the now cooled in pipe water between the heater and the faucet. A small inefficiency like that will lead to higher costs over time, and in many cases could be avoided through a better architectural design which places the water heater closest to the area that draws the most usage.

Bends in wiring will also occur, but should be avoided when possible for similar reasons. Bends will push more electrons into a smaller area, increasing heat. Heat, in turn, increases resistance and thus reduces current. In addition, the longer the current travels, the more resistance it will face, and therefore the more power loss you will see. This is true in CAT6 cables too, which transmit data through pushing electrons as binaries.

In either case, there are several basic rules to follow that will apply to data movement as well.

  1. KISS – Keep it simple, stupid. – If your pipes, wiring, or ETL/ELT pipelines look like a Rube Goldberg Machine, your process needs an overhaul.
  2. The closer the source and target are together, the less likely there is to be an issue and the less waste will be involved.

If either of the above is an issue, ask your data architect to get you a better design. Often the data engineers design overly complex pipelines because they are trying to band-aid broken business processes or because they are shoe-horning data into a suboptimal design. This is why teamwork is so critical. Data processes should support ALL users of the systems. If the source is jacked up, work with the source team to fix it. If the target is jacked up, then work with the data architect to fix it.

Leave a comment