Yes, we’ve passed the learning curve…
We started in 2012 researching and testing OpenStack. We’ve started and suffered with Essex. Now we are in Juno and implementing Ceph Giant.
We’ve passed through failures and amazing achievements. Failures caused most by the former project’s imperfections, the impulsion to test modules bringing more than instances’ orchestration, and a premature and innocent confidence on users’ advocacy to try better and newer services – we’ve faced some bad-behave customers renting instances to attack external sites and making us run to take care of it –
Failures make us stronger! Failures result in a serendipitous innovation.
Innovation runs into processes, people and culture.
Manage Cutting-Edge technologies like Openstack, OpenVSwitch and Ceph are really challenging. Install them and put them on production is just the starting point. It needs a skilled team, a motivated team to support them all time… sorry… not only support it, feel enough love for it to care of it… to fix it, code on it, keep it updated, adding constantly features, to keep it as attractive it was at the beginning.
This is the core, and it’s very reasonable to invest most time on it.
Invest time means get passionate people acting with autonomy: Millennials… an additional challenge.
Millennials are curious, demand vision and a defined cause to invest into; they get bored soon, unattached, like meritocracy and a straight honesty… culture to make them last in your organization…
Anyway, if supporting koolfit is not challenged enough, we have a world of things to do around!
So, now what!?
Getting “Staff On Demand”, an external attribute for ExOs (see my post: PUTTING PIECES TOGETHER BETWEEN INNOVATION AND EXPONENTIAL GROWTH), I’ve got the personal mission to work on our DevOps to avoid develop and operate services that we know we can get much cheaper and earlier using external providers. However, not any provider could apply, they have to bring APIs for automation to avoid any human interaction, bring an on-demand payment model, bring advanced features and enough flexibility to afford any change of the scope in our product. Meeting these conditions requires vendors working in a SaaS model. Well, this is the first filter to get the short list, later on, evaluate agility, commitment and competences… vendors offering a trial period are so much welcome.
Externalizing Monitor/Alert services
We’ve started building our monitor services on Nagios and later on Zabbix. However, we could see this project has got us out of our most important mission: maintain our core cloud stack stable and updated. We’ve realized build and monitor by ourselves would require time; people and we were so impatient to start yesterday!
We concluded after a couple of minutes of discussion we have to invest time searching for an application that help us to monitor our platform and customer’s applications. A SaaS provider with enough flexibility to integrate fast enough whatever we have now and later… serverdensity.com was our best candidate.
Next pictures are an extract of our monitor dashboard… You will know your availability metrics from inside and outside at any moment..
Get the required info from any node in your stack… virtual or physical…
Define thresholds and notifications as you wish…
Your web service is monitored from different locations around the world giving you more information about network issues …if they are related to your site… or is isolated to just one location out of your reach… other reasons maybe…
Of course, you monitor every openstacks component into your platform….
As we’ve done with monitor/alert services, we did with other applications for planing based on OKRs (Objective and Key Results), advanced DNS services, CDN, Anti-DDoS, etc….
Integrating ExOs into our operation will bring us more exponential attributes…
See you around!