Database
Some databases are configured with more capacity than they really need and that costs a
lot of money to maintain. We can take a look at your configuration and recommend whether
to downgrade compute or storage resources.
In an effort to keep a database management system in a maintainable state and keep the
number of security vulnerabilities low, cloud vendors tend to require an aggressive
regiment of upgrades in database software. Remaining on an old version for legacy system
compatibility often comes with an additional cost. Let us help you with routine
migrations to keep these costs low.
We can also take an inventory of your database schema and identify and remove any data
that is not being used.
Likewise, when data is no longer needed for
day-to-day operations, we can help you archive it and
even setup automated tasks for routine archival.
Logs
Every application handles logs in one of two ways: they either log way too much
information or not enough. In our experience, it is usually both at the same time.
Without careful planning during an application's design phases, developers don't really
know what should be logged and what will never be needed. This results in situations
where errors are happening all day long and never being logged, while in other parts of
the app, volumes of information is logged that no one will ever need.
We know how to analyze both scenarios and tame them so you will have only what you need,
when you need it. This not only makes applications far easier to maintain, it costs less
too.
See how we can use Grafana to manage logs efficiently.
Application
Most applications today are either underpowered or overpowered. We can examine usage
metrics and recommend configuration changes to scale up or down automatically as needed.
There are a lot more options available than what comes with default settings.
In some cases, this can be a long road, as optimizations for one service may not be
realized until an entire cluster can be scaled down to use fewer nodes. We can find out
if that applies to you.
If we host your applications for you, we will match your
services to the right infrastructure for maximum performance and maintainability. We
will also attempt to get the lowest cost. Many factors go into selecting the right
platforms, so the one with the absolute lowest cost might not be the best fit.
Metrics
Applications also tend to put a huge amount of numerical data in logs that can be
handled much more efficiently as metrics instead. By converting these log entries into
metrics, we can make the app much more observable with dashboards and alerts, and do it
at a fraction of the cost.
Keeping large amounts of metric data for a long time is another source of high costs. In
keeping with scheduled reporting needs, this data can be automatically summarized at
regular intervals, allowing you to keep more data for a longer time at a lower cost.
Managing your logs, metrics, and alerts efficiently paves the way for
effective process re-engineering, leading to further growth.
Alerts
In our experience, alert fatigue is the #1 reason why many technical issues never get
resolved. A system that behaves badly can generate so many alerts that everyone knows there
is no time to deal with them. This becomes more absurd when you look into some alerts and
find false positives or alerts for things that do not matter.
All of that noise drowns out the signals that really matter. This doesn't even count the
immesurable amount of time people waste on reading and deleting email alerts, not to mention
the compliance curiosities that surround unhandled critical alerts.
Ignoring the issue is not a good option. Let us audit all of the sources of alerts you have.
We can remove anything that does not matter. We can reconfigure ones that are good, but just
set for an unreasonable threshold. And for a bonus, we can even identify the false alarms
created by timing mismatches. Just because something happens every day at the same time does
not automatically mean it is an issue. It may just be a regularly scheduled batch job
performing critical business tasks. You'll want your alerts to take that into consideration.
What about the alert audience? Are alerts being sent to the right people? Or are too many
people getting alerts for issues they can't do anything about? Let us help you organize the
target audiences with group lists so everyone gets just the alerts they need, with none of
the noise they don't need.
It's an extra fun day when you get alerts that can't answer basic questions that would have
made them super helpful.
Take a look at some of the alerts you get every day or on special (unexpected) occasions.
Can your alerts explain basic information like:
- What app generated this alert?
- Did this happen in production or a test environment?
- What is the acceptable minimum or maximum value?
- What is the value right now?
- What value range have we typically been seeing for the past month?
- Does this tend to happen the same time every day?
- What are the likely causes that would make this happen?
If your alerts can answer questions like those, then they have what you need for fast,
effective incident response. If they don't, contact us today. We can teach your teams how to
write effective logs and use metrics to create alerts with actionable intelligence, leading
to a faster Mean Time To Recovery (MTTR) and a longer Mean Time Between Failure (MTBF).
