[ad_1]
Within the IoT world, a typical headache for purchasers is supporting their product after it’s launched. To supply assist with excellence, it’s vital to depend on the clever use of meta-support instruments to amplify the assist group’s work. On this publish, we’ll spotlight a few of these instruments.
Datadog is a wonderful monitoring service that may be prolonged and hooked into in some unbelievable methods. This method retains 24/7 surveillance on our cloud infrastructure, devoted hosts, web sites, and databases.
The options we use essentially the most are the screens. Displays could be easy smoke-test-style checks the place a instrument goes to a URL and makes positive it returns the anticipated HTTP code. Displays may also be advanced, fastidiously designed checks that cross enter to the instrument or a question to a database. For instance, now we have one verify that queries a Redshift database’s rows by time, averages the variety of entries made up to now hour, and returns a warning if the worth is beneath a specified worth.
Datadog has glorious first-class assist for all main cloud distributors, and if there isn’t a pre-made instrument, the platform is customizable sufficient to allow you to construct it your self!
In concept, this might be replicated by a grasp server pinging different servers or a free instrument (like the wonderful Cockpit Undertaking). Nonetheless, Datadog contains some distinctive sauce additions: its visualization of screens and metrics is unbelievable out of the field and has extremely granular instruments to, say, individually mute screens as soon as they’ve been triaged. When it’s essential to alert a assist group member, Datadog hooks into our slack channel to speak to the respective assist function.
Sure essential degree occasions additionally ship an automatic textual content to the on-call group member utilizing Twilio (extra on that later).
Chef
We have now a number of redundant Datadog monitoring cases, which we orchestrate and deploy robotically utilizing the Chef instrument.
Chef Software program is an in depth suite of instruments constructed round configuration administration. Paired with Datadog, this permits us to rapidly reply to points and maintain the whole lot we’re answerable for in a working, constantly reproducible state.
Whereas it could be a bit extra work the primary time to create a chef “cookbook” of recipes and use that to deploy infrastructure in comparison with simply doing it your self with a terminal, as soon as the work to make the infrastructure as code has been accomplished, additional work is way more trivial. Protecting our servers laid out in Chef cookbooks permits us to, for instance, apply Linux Kernel safety patches on a dev machine, robotically run assessments towards it, and when snug, apply the replace to each related machine and know what precisely has modified.
Chef additionally makes recovering from a catastrophe way more manageable. Ransomware? No downside, spin up a brand new EC2 occasion and execute your chef cookbooks! Mixed with different backup instruments, an entire server substitute is feasible with a easy invocation of the command chef-client.
The Chef neighborhood retains a public record of cookbooks up-to-date within the Chef “grocery store” to spherical issues off. For nearly any widespread activity you’ll be able to consider, anyone has a cookbook to assist implement it.
The one main draw back of Chef is the unlucky indisputable fact that these instruments share names with precise cooking and purchasing phrases, which might take a look at the abilities of any developer’s Google-fu. Instance: Chef has a command-line instrument for managing dependencies referred to as “knife.” In case you Google “chef knife not working,” you get many outcomes, only a few of which is able to assist you to arrange a server.
Luckily for us, now we have discovered Chef’s documentation, supplemented with a wholesome seasoning of StackOverflow, solutions 99.9% of questions.
AWS – CodeCommit/CodeBuild
Itemizing AWS as a meta-tool is a bit like itemizing “a pc” as a developer’s tech stack. It’s technically correct however, effectively, a bit normal. AWS can do quite a bit (ask anybody who has studied for one among their exams), however as we speak I’ll deal with the bits we use particularly for our assist pipeline: CodeCommit and CodeBuild.
CodeCommit isn’t too thrilling by itself: it’s a personal git host. Its usefulness comes, as with many AWS merchandise, within the tight integration to different AWS instruments. Storing code in CodeCommit permits us to set off CloudWatch occasions on commits with sure messages and run computerized linting, CI/CD, and execute builds. We aren’t too fancy; our assist group has a pleasant however simple workflow the place we maintain all of our Chef cookbooks as particular person CodeCommit repositories. We will then use CodeBuild to supply artifacts. The chef infrastructure then checks every day for any updates and applies them to all our related machines.
Miscellaneous
To date, this weblog publish has walked via the first “loop” of assist: infrastructure, coded for Chef, deployed with AWS, and monitored by Datadog. In fact, these aren’t all of the instruments, however I sadly don’t have the time to write down (nor most of our expensive readers, to devour) the handfuls of pages it could take to fill in each element. As a substitute, I’ll briefly point out among the different instruments exterior of this major assist pipeline.
CrashPlan
For some legacy programs and instruments, CrashPlan is a dead-simple automated system backup instrument. It could carry out an entire system restoration, however we frequently discover ourselves utilizing it to assist a buyer get well a person modified or deleted file. It’s a useful gizmo that you just hope by no means to make use of, however boy will you be glad when it’s important to use it.
Twilio
To make sure we are able to reply to emergencies at any time of the day, now we have Datadog hooked right into a Twilio connection, permitting us to textual content the on-call assist group member underneath sure circumstances. On this occasion, Twilio is simply the pipe; Datadog and a few cron jobs carry out the brunt of the work, however with out the texts, the assist response would probably be missing exterior enterprise hours.
Confluence
Realistically, you possibly can put all kinds of wiki instruments on this bullet. Atlassian’s Confluence product is polished, has each potential knob and button you possibly can think about, and ties in with different Atlassian instruments. Documenting our options and pipeline within the wiki is essential for elevating our group’s “bus issue” — the hypothetical quantity of people that might, for one motive or one other, disappear (be “hit by a bus”) with out our assist struggling.
The Future
Lengthy-term, our aim is to make the assist function as pointless as potential. Whereas we perceive {that a} human assist specialist function is all the time going to be wanted, by magnifying the effectiveness of our group, we keep away from burnout and wasted time on micro-managing points. As a substitute, we are able to deal with the vital a part of our work: bringing our prospects’ concepts and options to fruition with IoT!
We’re wanting into some thrilling and promising instruments, together with Ansible as a possible complement or substitute for Chef, changing all guide SSL cert renewal with computerized renewal through an ACME-client registrar, and hybrid-cloud options utilizing Terraform to permit cross-cloud fallback.
Conclusion
We hope this brief dive into assist tooling evokes or encourages you in your assist endeavors. As an business, serving to others improve their high quality and longevity of assist for IoT services and products in the end helps everybody, prospects and firms alike.
[ad_2]
