Start trial

    Start trial

      The planet has been warming at a steady pace and it's projected to keep doing so. While technology has played a part in this situation, it can become part of the solution.

      Earth is warming at the fastest rate in our history, and most scientists now agree that humans are the main contributing factor1. Since the beginning of the Industrial Age, human activity has been responsible for an increase of 45% of carbon dioxide, the most common of the heat-trapping greenhouse gases, in the atmosphere2. It is no wonder that the last 5 years have been the warmest in centuries3.

      The Information Age has delivered a colossal rise in the number of computer processors, which is the single biggest energy consumer in any computer system or smart device.


      A sharp increase in an already steep curve

      Whilst we need greenhouse gases in moderation, this sharp increase is creating an aerial blanket. These gases absorb solar energy and keep heat closer to the planet surface, whereas in the past heat would have escaped into space. This is causing more frequent extremes in weather and global climate change is threatening our very survival.

      As global citizens we all have a part to play. As individuals and as Industries, we all need to reduce our carbon emissions now and adopt a more sustainable way of using energy in the environment.

      What part technology plays in this picture

      Technology has a big part to play in decarbonizing the planet. The Information Age, which started in the seventies, has delivered a colossal rise in the number of computer processors. While there are now estimated to be more than 11 billion devices connected to the internet4, almost 3 billion people worldwide do not have access to this technology5.

      The processor is the single biggest energy consumer in any computer system or smart device. The speed of the processor is strongly tied to its power consumption. Over the decades, as chip manufacturers have delivered significantly faster processor speeds, their electrical power consumption has also increased. Observations of computations per joule of energy in technology since the 1950s has shown a large drop in efficiency over the last 20 years6. This suggests that compute power became the priority over energy efficiency.

      CPU vendors rate their chips according to clock speed in MHz/GHz - higher clock speeds require more power and generate more heat. Every computer device expends heat in proportion to the electricity it consumes. This is the reasons mobile devices like smart phones tend to run at lower clock speeds than larger computer systems.

      The amount of heat that a computer produces varies in relation to the amount of work it does - the busier the processor is, the more heat it produces. Our goal must be to decarbonize and reduce compute.

      How can technology turn it around?

      img-computer-chip-and-computer-circuitry-with-green-leaf-01Sustainable compute solutions include living on and working from buildings powered by green energy, with net zero or negative carbon emissions, but what can we do in the technology industry to further help?

      Migrations are a common activity for most technologists. They usually involve identifying the main drivers/goals for the change - technologist should use this as an opportunity to steer the migration to a greener direction:

      • Most computer system vendors publish their kW/s consumption for the product. If looking at migrating to new hardware, then compare the figures between the current and new to ensure the new hardware is more energy-efficient.
      • One of the goals in any migration is to consider consolidation opportunities. Often, non-production environments are the easiest to target, as not all of them need to be production-like in their performance. Consolidation opportunities could involve database instance or server sharing.
      • Often, non-production environments (and sometimes even production environments) contain servers that are not used outside core hours, and yet they are powered 24x7. Consider host shutdowns when not in use.
      • Reduce energy consumption, decrease energy cost, and provide maintenance windows by having an automated schedule to power off SaaS instances at the end of core hours and then power them back afterwards.
      • Consolidation may result in reduced compute needed. Clock speeds and cores should be evaluated and tested, as sometimes chips showing slower clock speeds may have a faster burst, and this could reduce burning energy.
      • Software refreshes during migrations, allow for OS and DB security patching opportunities, whereas legacy systems may have been out of support.
      • Check if the new hardware uses recycled materials, and if so, what percentage that was, for comparisons.
      • Check if your “old” hardware parts can be recycled.
      • If migrating to a new data center or cloud solution, ask about the building's energy efficiency rating. The higher the rating, the more energy efficient the building is. Some data centers are carbon-negative, generating more green energy than they consume.
      • Asking more eco-friendly questions about the data center efficiency and eco-opportunities will help make people take note.

      Fujitsu Enterprise Postgres greener operations


      Migrating the current RDBMS to Fujitsu Enterprise Postgres could provide opportunities to reduce compute demand. If we can reduce the compute demand, we may be able to reduce CPUs or cores, which would reduce the carbon footprint. The next points are greener check list items for review.

      • Reducing the demand on compute would first ideally involve analysis of the current performance and architecture schematics.
        • It could include increasing RAM, increasing IOPS to improve the execution times and overall load on the system.
      • Identify SQL with high execution counts to potentially reduce the processor demand with strategic materialized views optimization.
      • Identify SQL aggregation which would be more efficiently done using Fujitsu's Vertical Clustered Index.
      • Review long-running SQL for more efficient query plans and improved index optimization.
      • Identify SQL scanning on large tables that could benefit from new indexing.
      • Identify SQL scanning on large tables that could benefit from table partitioning.
      • Review SQL to use “current day” key values in lookup tables. Most business OLTP systems tend to only be interested in recent data whilst storing much older data. When query plans regress, it means that more processing is needed.
      • Review client connection flows and connection pooling opportunities. Fewer concurrent connections will reduce the system load. High system loads mean higher processing.
      • Review idle connection timeouts and query timeouts. If these aren’t managed, they can also add unnecessary system load.
      • Review opportunity to use Global Meta Cache to reduce query processing time.
      • Align the OS mountpoint block sizes with the Fujitsu Enterprise Postgres block size. Misalignment can cause additional IO.

      Post-migration considerations for greener compute


      After taking into account the factors above during migration, there is even more that you can do to decarbonize your compute. Keep an eye on the following:

      • Energy savings - Power off instances when not in use outside core service hours.
      • Health checks - Malware, viruses, and rogue processes can all cause high CPU use, which will use more energy than when CPU is idle.
      • Horizontal scaling – If on a distributed database, consider offloading read operations to read-only-replicas and only perform writes on the primary. This should improve concurrency and reduce waits/retries.
      • Purge/archive old data – Unless required, look at archiving old data and removing it from the online database.
      • Perform regular housekeeping
        • Check for unused indexes that could be removed.
        • Check for SQL that could run more efficiently with indices.
        • Schedule regular vacuums.
        • Schedule regular index statistics analysis.
      • Investigate Materialized View opportunities to reduce repeat compute.
      • Consider Vertical Indexes where aggregation would reduce compute.
      • Consider connection pooling to reuse connections.
      • Review idle connection and query timeout configuration to reduce unnecessary orphaned sessions.
      • Consider using Global Meta Cache to reduce query processing time.
      • Compare the existing encryption solution against Fujitsu Enterprise Postgres' Transparent Data Encryption to check if it is as CPU-efficient and easy to operate.

      It's in our hands to make the change

      These are just some of the ways in which we can make a positive change and contribute reducing our impact on the environment.

      If we all help reduce carbon emissions, we can collectively make a difference.


      Sources used in this article







      Subscribe to be notified of future blog posts
      If you would like to be notified of my next blog posts and other PostgreSQL-related articles, fill the form here.

      Topics: PostgreSQL, PostgreSQL community

      Receive our blog

      Fill the form to receive notifications of future posts

      Search by topic

      see all >

      Read our latest blogs

      Read our most recent articles regarding all aspects of PostgreSQL and Fujitsu Enterprise Postgres.

      Receive our blog

      Fill the form to receive notifications of future posts

      Search by topic

      see all >