SLA is an acronym for Service Level Agreement. In the popular mind this is all about promised up-time versus actual up-time from your service provider for an api, a host, application, database or resource. It's a popular tech buzzword but it isn't always an indicator of positive service results...

IT pros and customers alike will throw around terms like "the five nines" as the perfect gold standard for SLA metrics. This means that for any particular service, that service will be available 99.999% per year. This equates less downtime of 1 second per day, 6 seconds per week, 26 seconds per month and finally 5 minutes and 15 seconds per year unavailability. Some SLA's do not include scheduled downtime permitting planned periods of maintenance and upgrades. This is where the slippery slope begins and Marketeers usually pick up the sales-call conversation.

So what does this really mean for a customer?

Well context is key in this type of discussion. If a SLA includes planned downtime, that might be meaningless - unless customers aren't aware - or have been communicated with responsibly. From a certain perspective that is normal operations while others would claim this type of agreement exhibits poorly executed planning.

Perhaps a degraded experience might be acceptable that won't trigger a knock to the SLA agreement. In the real world, a maintenance period might be performed at 3AM EST or on a Friday at 8pm PST while demand is essentially zero with no adverse effects. Alternatively, it's common for businesses to disregard certain holidays as "non-work" when IT support teams and developers are permitted to rollout features from their sandboxes into production status after extensive testing.

Not everything can go perfectly smooth without flaws for we too in information technology are human after all like the rest of you, and can make mistakes. Sometimes its just impossible to predict the future no matter how good the plans are.

Large providers, your Microsofts and Ciscos of the world, who support vast domestic and international economics at mindbogglingly scale (and responsibility) are keen to work even beyond the "five nines" with good reason. Having billions in financial resources makes this possible.

Systems are built to be resilient with multiple redundancies, regional routing, high availability, can scale up and scale down dynamically, utilize substantial caching (end users might not even notice) or users can continue to work locally, semi-offline (and invisibly) with Service Workers. But these giants are also not immune as any System Administrator can tell you.

Most large providers have extensive systems in place for up-time status notifications, monitoring, dedicated live backups and highly skilled response teams to run infrastructure (teams larger than most average small businesses) for when things go bad. Somethings are out of control of providers: underseas cables get messed up by fishing trawlers, power companies like PG&E accidentally burn down cities and infrastructure, and yes, nerds simply goof up DNS really, really badly (looking at you Cloudflare).

For the small business owner, downtime will usually equal money loss. Especially unplanned. We encourage small businesses to support those other small businesses in the tech space, not just the giants of industry, because leaner can be better and offer pivoting advantages. Most everything the tech giants want to charge you for hundreds and thousands a month with a pervasive (blackmailing) SaaS model can be done by open source software that respects privacy and liberty - for substantially less.

We've written in the past how we care for our customer's servers like our own. Now we've decided to put our money where our mouth is by sharing some current and historical SLA data with you our visitor, blog reader and potential client.

One year and all time up-time for four web hosts.

The image above shows four web server hosts we currently service with stats for 1 year and "all time" since we started recording data (circa 2018). Two are our own sites and two are clients' websites. The former represent this website and a very small portfolio site (really just a demo of minimalist design).

The latter websites represent one company rated at around $25 million per year and the second company around $2 million annually. They both are B2B models and rely on extensive rep agencies, operate in the design-build construction industries to sell products to architects, lighting designers, landscape architects, specifiers, etc. To put it bluntly, their websites, aside from legacy sales relationships, is their number one method of funneling income into the business. Website availability is very critical to serve approximately 300 reps and 100 internal employees.

Our data, due to thousandths precision rounding, puts our hosting service here at Grip Fast Information Services & Technology into the "the four nines" level of up-time SLA. This is done for less than $100 per month in actual hosting costs across all 4 websites. This period also includes all maintenance time, upgrades, project-phase roll-outs, service expansions, etc all while continuing to serve customers, process email, data requests without skipping a beat. Further, these sites have been performing at this level (with data to back up the claim) for over 5 years at the time of this article publishing.

For some further context, these four websites have served (from March 1st 2022 through March 2nd 2023) 158,942 visits equating 575,209 page views with average unique visitor count around 3000 per month - all with privacy respecting analytics via Matomo. These are custom built, managed hosts with original and highly customized websites costing far less and performing insanely better over offers from Squarespace, GoDaddy, Wix, ad nauseum.

Please note: We currently compile the up-time data via GitHub Actions. Though this system is simple, robust and very economical there is some variance in the reported data - as there is no assurance on scheduled jobs at Github. This means that our up-time report data in reality is higher than what is being reported. In simple terms, our SLA percentage report is lower than the actual performance by a full decimal place putting our hosting beyond the "four and half nines" category (again a fraction of costs to our clients over previous vendors).

Has this peaked any interest in migrating your clone website and slow server to something better with high availability from a local partner not some faceless mega-corp? Contact us! We are your technology experts!



Comments powered by CComment

Member of The Internet Defense LeagueOpen Source Initiative