I'm not sure my brain is up to wading through the site trying to figure out what this is. Can someone break this down into it's purpose and components? Is it a code deployment system, container manager, open-stack, chef/puppet/fabric?
It's a multi-cloud application orchestrator. OneOps lets you design your application in a cloud agnostic way (by abstracting multiple cloud providers). It manages your application's design, deployments, operations & monitoring.
At the moment these cloud providers are supported - http://oneops.com/integrations.html#clouds
It's a multi-cloud application orchestrator. OneOps lets you design your application in a cloud agnostic way (by abstracting multiple cloud providers). It manages your application's design, deployments, operations & monitoring. At the moment these cloud providers are supported - http://oneops.com/integrations.html#clouds
"Accelerating product delivery in the Digital Economy through Continuous Application Lifecycle Management of Cloud-based Workload, backed by @WalmartLabs"
I think you could work on that elevator pitch a little - it seems like it's been workshopped by a committee, and doesn't really tell me anything.
You might want to brush up on business speak, that's the real reason for writing the vast majority of all software ever written or that will be written in this century.
The point is that it is possible to be able to pitch a technology to management without being overly opaque and sounding like a bag of buzzwords.
Also, keep in mind that different industries are accustomed to different sets of terminology. Engineers at tech companies don't usually use terms such as "product delivery" and "Continuous Application Lifecycle Management." To us, "multi-cloud orchestrator" is a much clearer way to describe this technology.
It's odd, I am an engineer at a tech company that works on the Salesforce stack so I use the phrase "product delivery" all the time and "continuous application lifecycle management" is a perfectly normal phrase to me. It's really all about different industries, even within "tech".
Not only that, this is coming out of "Walmart Labs @ Walmart Global eCommerce". Sounds like a tech company.... This is more just big company PR people than tech or non-tech.
Its a Multi-Cloud Orchestrator. It lets you abstract multiple cloud providers, so you can say use Azure, S3 , Rackspace Openstack, your own openstack and your own VMware all at the same time, and have unified management.
At the moment it supports any cloud with a OpenStack endpoint/integration.
Thank you. Now that you've summarized this, this sounds like a great idea, a good way to kill lock-in on cloud platforms. I don't know why they couldn't write something so simple and concise...
Does that not mean that you can then only use the common aspects from the various cloud providers, and miss out on the additional features each supports?
That's the difference between startup and big environment. Bells and whistles are nice, but handcuff you.
If you need a 10 year business plan for a service, you need to limit your choices. I have systems with 40 year history that need to retain data for 30 year into the future. Different thinking.
- When you say "40 year history", do you mean the system was originally written in the 70's, or do you mean it has data going back to then?
- Why do you need to keep data for 30 years? Regulatory reasons? Reporting? Or just long-lived data?
I'm not doubting anything you're saying. It's easy to imagine systems with those attributes, say, in banking, insurance, legal records, etc. I'm just curious about your specific case. :)
Long after the logic of your system has been replaced, the data will live on. It's something I think people tend to undervalue.
No problem -- I'm really passionate about this topic!
.gov stuff. Every state in the union has social services systems that were stood up in the 60s and 70s, and have or are being transitioned over time from whatever mainframe they were in to modern stuff.
Usually the long term stuff is retained. These folks deal with situations like disabled children who become wards of the state. Their records need to exist well until those children are in their 40s.
Thats just one example -- Public works projects are built with 50+ year lifespans, and many of those records are expected to exist for at least that long.
There's also a whole historical responsibility thing that I think will become more important. You can develop a deep understanding of how a governor or Mayor made certain decisions going back to the American revolution (my county has records of the legislative body from the Dutch colonial period). Digitalization has erased a lot of that -- in many cases we cannot read data from the 1980s.
I think issues like this are already starting to appear on a small/personal scale -- when my grandfather died in 1980s, all of his business records, personal records, pictures, etc were on paper somewhere in his home. You could sift through and keep important things. What do you do when the only protection your data in Dropbox/et al is one missed credit card transaction away from oblivion?
BTW, I think you're 100% correct about this becoming more important / more of a visible problem. On a small scale, it's something I wrestle with often. For example, I love the convenience of buying a book on my Kindle, but I'm reminded of the joy of wandering through my grandparents' library and wonder if my daughter (and so on) will be able to have a similar experience – what happens to those my Amazon purchases after I'm gone? Another example is that I keep a sporadic journal in Evernote. Should I really be printing out those writings in case I have a curious descendant? I've enjoyed reading letters and journals of ancestors.
On the other hand, digitalization has obvious benefits too. But we have to be deliberate in our curation and storage format.
For example, it's great that correspondence (email) is now automatically archived rather than it being a tedious process (carbon copy, hand-copying, etc.). But how many people backup their email out of their cloud provider to a personally owned copy in a human-readable format? I actually do backup my Gmail account occasionally. But I doubt my wife knows where it is. Or if she'd be able to look past the XML format. And even if she did, there are literally millions of emails in there from mailing lists and work stuff and spam and whatever. The percentage of actual meaningful correspondence is unbelievably low.
So their is no standard on how to present your cloud to the world ATM. I think you would be happy that widely used open-source project like OpenStacks interface is being adopted.
I'm sure if you want OneOps to support a non-OpenStack interface that should be possible too, they already support Microsoft Azure for example.
Thanks, they really need to write a simple summary at the very top that anyone could understand. I guess it's really their first time trying to share something (as far as I know) so they must be accustomed to speaking business lingo.
I actually have no idea. After posting that, I looked at deltacloud git, and there haven't been any new commits since 2013, so it doesn't seem too active.
As your business grows at some point it may be cheaper to run your own cloud (OpenStack or something else), and you can start migrating your workloads with a PaaS layer like this.
At my previous employer the loaded TCO for running in-house OpenStack was 45-50% of AWS (the company used both).
"Loaded TCO" means gear, space, power, cooling, transit, people... Those are high-level, and I don't remember a bunch of things that went into the 5-year TCO model but it was complex.
If you have the talent and scale it's certainly doable, and like I said, we were at less than half of AWS. There are plenty of examples, one that comes to mind that is NOT hyper-scale is Server Fault.
IMO, it is more than that. It allows you to model your application (currently supports these software stacks http://oneops.com/integrations.html#software) in a cloud agnostic way and manages it's life-cycle
So walmart thought this startup was worth buying to manage all their different applications running in different clouds. Fast forward a couple years and they've now open sourced the entire product. That's pretty cool.
You might want to take a look at https://github.com/StackStorm/st2 if you want python support, chatops, UI, a GUI editor for writing workflows and a strong community.
Disclaimer: I am a programmer at StackStorm. I'd be happy to help.
Fair enough. +1 to ChatOps enabled via StackStorm. Disclosure - I work at StackStorm. What we have seen is that over time event driven automation can play the role of any such opinionated approach to cloud orchestration and even as an alternative to heavy reliance on PAAS layers. Event driven automation like StackStorm tends to err on the side of flexibility (use your own scripts, click together the workflows, do your own IFTTT) whereas solutions like OneOps are more polished out of the box and more focused on a particular pipeline. Having said that - StackStorm for ChatOps and especially remediation is the most common way people adopt StackStorm and that use case is quite different than the Walmart project.
I'm kinda surprised by this. My impression was that Walmart Labs had great success in deploying a pure-openstack private cloud in record time which was immense in scale. Why would they want to deploy to other clouds?
Since Amazon is basically Walmart without brick and mortar and they make some compelling money from their AWS, why wouldn't Walmart be interested in getting in to that business?
Has anyone seen anything talking about resource requirements for deploying this? The fact that Cassandra, ElasticSearch, PostgreSQL are all involved plus much more makes me think the requirements must be quite high.
The chef code that deploys all the apps is pretty old skool, copy pasta'd from opensource cookbooks in alot of places, will never pass food critic or rubocop, and has no tests. I wouldn't really want to run that code on anything.
Because opensource cookbooks are unreliable or people should only rely on your cookbooks? Or old skool isn't new skool so it should be wholly disregarded?
open source or not, they should follow the best practices, these don't.
The fact is, really old open source cookbooks have been copy pasted, with the maintainers names changed, and hacked on top of. For example, the MySQL cookbook. This is against the apache2 licence that they were originally distributed under.
It's great to see established companies experiment with OSS. The issue I see is that they make their projects go live/public a little bit late to be usable 'as is'. By the time they open source their tooling, the stack maybe out of synch with the latest and greatest. Just the fact OneOps is using nagios checks for monitoring tells you a lot.
81 comments
[ 6.3 ms ] story [ 65.2 ms ] threadhttp://oneops.github.io/admin/key-concepts/#oneops-system-ar...
Side note, every-time I see an enterprise message bus, I throw up in my mouth a little.
It's the cloud equivalent of a Super Wal*Mart
Looks fairly complete and extensible, though.
And by the way, it's not just OpenStack. Looks like it's also tooled for AWS.
I think you could work on that elevator pitch a little - it seems like it's been workshopped by a committee, and doesn't really tell me anything.
Also, keep in mind that different industries are accustomed to different sets of terminology. Engineers at tech companies don't usually use terms such as "product delivery" and "Continuous Application Lifecycle Management." To us, "multi-cloud orchestrator" is a much clearer way to describe this technology.
No need to be so condescending.
At the moment it supports any cloud with a OpenStack endpoint/integration.
Blog post: http://www.walmartlabs.com/2016/01/oneops-now-available/
If you need a 10 year business plan for a service, you need to limit your choices. I have systems with 40 year history that need to retain data for 30 year into the future. Different thinking.
- What types of systems are these?
- When you say "40 year history", do you mean the system was originally written in the 70's, or do you mean it has data going back to then?
- Why do you need to keep data for 30 years? Regulatory reasons? Reporting? Or just long-lived data?
I'm not doubting anything you're saying. It's easy to imagine systems with those attributes, say, in banking, insurance, legal records, etc. I'm just curious about your specific case. :)
Long after the logic of your system has been replaced, the data will live on. It's something I think people tend to undervalue.
.gov stuff. Every state in the union has social services systems that were stood up in the 60s and 70s, and have or are being transitioned over time from whatever mainframe they were in to modern stuff.
Usually the long term stuff is retained. These folks deal with situations like disabled children who become wards of the state. Their records need to exist well until those children are in their 40s.
Thats just one example -- Public works projects are built with 50+ year lifespans, and many of those records are expected to exist for at least that long.
There's also a whole historical responsibility thing that I think will become more important. You can develop a deep understanding of how a governor or Mayor made certain decisions going back to the American revolution (my county has records of the legislative body from the Dutch colonial period). Digitalization has erased a lot of that -- in many cases we cannot read data from the 1980s.
I think issues like this are already starting to appear on a small/personal scale -- when my grandfather died in 1980s, all of his business records, personal records, pictures, etc were on paper somewhere in his home. You could sift through and keep important things. What do you do when the only protection your data in Dropbox/et al is one missed credit card transaction away from oblivion?
BTW, I think you're 100% correct about this becoming more important / more of a visible problem. On a small scale, it's something I wrestle with often. For example, I love the convenience of buying a book on my Kindle, but I'm reminded of the joy of wandering through my grandparents' library and wonder if my daughter (and so on) will be able to have a similar experience – what happens to those my Amazon purchases after I'm gone? Another example is that I keep a sporadic journal in Evernote. Should I really be printing out those writings in case I have a curious descendant? I've enjoyed reading letters and journals of ancestors.
On the other hand, digitalization has obvious benefits too. But we have to be deliberate in our curation and storage format.
For example, it's great that correspondence (email) is now automatically archived rather than it being a tedious process (carbon copy, hand-copying, etc.). But how many people backup their email out of their cloud provider to a personally owned copy in a human-readable format? I actually do backup my Gmail account occasionally. But I doubt my wife knows where it is. Or if she'd be able to look past the XML format. And even if she did, there are literally millions of emails in there from mailing lists and work stuff and spam and whatever. The percentage of actual meaningful correspondence is unbelievably low.
I don't know. It's a big problem, I think.
I'm sure if you want OneOps to support a non-OpenStack interface that should be possible too, they already support Microsoft Azure for example.
https://i.imgur.com/wxSi05R.png
At my previous employer the loaded TCO for running in-house OpenStack was 45-50% of AWS (the company used both).
If you have the talent and scale it's certainly doable, and like I said, we were at less than half of AWS. There are plenty of examples, one that comes to mind that is NOT hyper-scale is Server Fault.
Disclaimer: I am a programmer at StackStorm. I'd be happy to help.
Edit: Full disclosure, I am the founder of ContainerShip.
https://www.openstack.org/summit/vancouver-2015/summit-video...
* cloud bursting during high load
* disaster recovery
* putting things in geographic regions where they don't have their own hardware
Since Amazon is basically Walmart without brick and mortar and they make some compelling money from their AWS, why wouldn't Walmart be interested in getting in to that business?
You might also want to quickly gain a presence in a geographical region you don't have hardware in (ever or yet).
Christmas / Thanksgiving shopping in the US would be my guesses.
Has anyone seen anything talking about resource requirements for deploying this? The fact that Cassandra, ElasticSearch, PostgreSQL are all involved plus much more makes me think the requirements must be quite high.
Edit: nevermind http://oneops.github.io/admin/prerequisites/
Why, for the love of God, why?
They should have put up a blog post with an overview like this one: http://techblog.netflix.com/2015/11/global-continuous-delive...