InteropCloudChat

OpenStack seems to have gone in two directions: [1] focused on Telco/NFV and [2] delivering hosted OpenStack

I keep hearing about issues with the networking component

@marciasavage yes, OpenStack networking is still considered a challenge. What's the latest @mestery?

Difficult question. With the exception of networking on paper it offers technology required. I think brand and brand awareness around OpenStack is a problem. One of it's failings may be marketing of all things.

@CTOAdvisor agreed, and it's often confusing about where OpenStack should stop and OpenDaylight should start in networking.

We are going to have the Infrastructure Director from Walmart labs talking about how the company has doubled down on OpenStack at interop. I'm really interested to hear what he has to say

@SusanFogarty I'm interested as well. So many questions around culture acceptance in addition to the technology challenges.

@SusanFogarty Me too. We just posted an article about Walmart's OneOps platform. Interesting project.

@CTOAdvisor apparently they have a corporate mandate to become more technology-centric and go open source, and the IT folks are all on board

My company definitely leans heavily toward productivity apps as far as SaaS goes. Many of those were brought in by specific users (guessing that's not uncommon)

once they are in place, are they managed by each group, or do they get centrally managed?

Almost all our use of SaaS is productivity focused. Return question: Are we including infra services delivered as a service in the category of SaaS? Or do they need a new category?

@lmacvittie I don't think the infra-SaaS apps get enough attention. great way to augment IT skills.

@lmacvittie interesting point. What should we call infra SaaS?

@SusanFogarty ISaaS?

Big trend I see is internal systems moving to SaaS. All the non-sexy stuff that is needed to run biz internal

.@aarondelp Exactly! Biz ops moving to SaaS. Anything biz fn or process that is commoditized is headed for the cloud.

@aarondelp agreed. none of those systems differentiate a business like they did back in the 90s or '00s.

@lmacvittie what are some of your favorite iSaaS applications?

Most are security related right now. DDoS protection, CASB. Waiting on ID federation/SSO... more of that, please. Too many identities ;-)

Course DNS,too. The original iSaaS, right?

@lmacvittie would you include the SaaS apps focus on logging, cost-management, ChatOps in iSaaS?

Logging, yes. ChatOps though, that's more process management. Ops processes, but process. Like BPM for ops. ;-) I guess it could, hadn't really thought of it that way....business infrastructure I'd call that. Huh. Interesting notion.

Every company has a window where public cloud makes sense but that window has a top end where the utility model becomes too expensive. Ops are another story though...

Not unique or the first. We'll see others as they hit that tipping point where costs to rent are more than costs to build.

@lmacvittie Even with AWS always cutting its prices?

@marciasavage Yup. It isn't just about the cost of compute. Bandwidth and other services have to be figured in. They're rarely cut as deeply as compute.

One of my first thoughts was physics. How do you migrate 500PB of data. Next question was cost. How did they successfully negotiate the exit and was it part of their existing contract. The data migration alone is extremely expensive.

@lmacvittie @aarondelp care to take a WAG at what the tipping point might be?

@CTOAdvisor agree that it would be incredibly expensive to exit all the data, but their service is paying bandwidth costs for every file sync. networking costs had to be one of their largest costs.

The move doesn't surprise me as much as the logistics and lessons learnt. How can enterprises stay agile enough to move from public to private successfully. It helps that Dropbox basically has one service.

I think it's when the runrate for operating in the cloud starts eating into other ops (and eventually margins). There's got to be a formula for that but that'd require more biz expertise than I have.

You bring up an interesting point. Do you reach a point where the monthly costs are hurting but getting the data out will completely kill you in transfer fees. Interesting thought...

@aarondelp Do they not have the reverse of their "ship us your data and we'll get it loaded" service? Cause that's how you get gobs of it *into* the cloud in the first place.. not on the wire.

@lmacvittie Great question, I have never seen one but can't answer for sure

@lmacvittie yes, AWS had a "send us your disks" service, and now it's packaged in a fancy box with a Kindle for tracking, called "Snowball".

@aarondelp I've not seen one either. Assuming none, your point re: costs too high but data out will kill you is a good one. What do you do then?

@aarondelp @lmacvittie and that service now supports "we'll send your data back to you". I believe it's in 50TB chunks.

Nice. That helps.

We were 30 minutes into the chat and nobody had mentioned containers, so it had to be done. Drink!!

Nobody wants to answer this so I will: yes, and dev/test.

Docker usage stats from DataDog: https://www.datadogh...

Docker usage stats from NewRelic: https://blog.newreli...

DevOps.com Docker usage stats: http://devops.com/20...

LOL, I don't honestly know the answer to this question, but I'm guessing test/dev

StackEngine Docker usage stats: http://www.stackengi...

Some very differing results of those surveys. Looking forward to seeing @dberkholz upcoming data

What do you make of the different results?

@SusanFogarty I think lots of people have Docker on their laptops or Dev/Test. the big %s tend to be skewed by people that want to answer the questions. the "resume-driven development" phenomenon.

Yes, makes sense. And we see that bias a lot at well

In our enterprise surveys, we are consistently seeing around 15% of IT shops w/ containers in prod, around 35% that have reached pilots or beyond.