Up next in 10
The Cloud Show with Magnus Mårtensson ft. Stephen Thair - Ep. 5
8K views
Oct 30, 2023
The show is about cloud leadership and all the important questions relating to cloud projects. Certainly, many matters when a company is going to and want to be successful in the cloud, are about technology. However, there are many additional matters, adjacent to technology, that we also need to tend to regarding business strategy, human resources, organizational change, planning for a technical cloud approach, and many more questions. These conversations are critical for a healthy cloud and for a swift and accurate cloud approach. GUEST SPEAKER Mustafa Toroman: Experienced Solution Architect with a demonstrated history of working in the computer software industry. Skilled in Cloud Computing, Microsoft Azure, MS SQL Server, Data Migration, DevOps and Public Speaking. Microsoft Certified Trainer and Microsoft MVP for Microsoft Azure. C# Corner MVP and award wining book author. C# Corner - Community of Software and Data Developers https://www.c-sharpcorner.com #CSharpCorner #TheCloudShow #CSharpTV
View Video Transcript
0:00
down or hello hello and welcome to the cloud show i am magnus martensen and we're back we're back
0:08
with another interesting topic it's a complicated one but it's very very important to cover this
0:15
because people will come and say hey how about we be on more than one cloud i mean we're such a big
0:20
company we can't just put all of our eggs in one cloud basket no no no no we have to use more than
0:26
one cloud because otherwise bad things will happen. Is that true or is that just something
0:32
we need to converse about and see if there's another solution maybe? The guest of the show
0:37
today is going to be Steve Thayer. He's one of my good friends and he's a really, really
0:41
experienced guy when it comes to these things. So welcome to the cloud show
0:56
Hello, my friend. How are you? I'm very well, Magnus. How are you
1:00
I'm so good. I'm so good. It's nice to be back here on the show and talk about a topic which is so interesting
1:09
I've had this conversation with many customers of mine that felt that
1:15
oh, we're such a big company. We're huge. We're this enterprise and all these things
1:19
So, of course, we can't just use one cloud. So they deliberately choose to use more than one cloud
1:24
And to differentiate, we're not talking about hybrid here when you have some things on premise
1:30
That's the reality for many. And then you start using the cloud, so you kind of extend to the cloud
1:36
No, no, no. They're deliberately doing this. They're saying, we need to use more than one cloud
1:41
And I know you have some thoughts on this topic. Oh, I have many thoughts on this topic
1:45
Unfortunately, we've already got 20 minutes. So I think, first of all, so my name is Steve Thayer
1:52
I'm a former CTO of DevOps Group, ran a DevOps and cloud company for 10 years, been in the industry 30 years
1:57
I've been in RD for five or six years. I think we've really got to define what we mean by multi-cloud or a multi-cloud application
2:06
And I did a little, I did a very quick survey on LinkedIn. I only got 55 responses, so it's not super accurate
2:12
But, you know, it came back, the number of people running the same workload on multiple clouds was only 11%
2:21
Okay, so that's really my true definition of multi-cloud, a single application running on multiple clouds for whatever reasons that you need to do that
2:33
Different workloads per cloud, 56%. Right. You know, all in on a single cloud, 27% and other about 5%
2:44
Yeah. And I think it's super, super interesting to, you know, you and I both work with, you know
2:50
often work with very large enterprise organisations. Every enterprise organisation that I've dealt with is multi-cloud
2:56
Not necessarily because they chose it. It's because one part of the business went one cloud
3:01
and one part of the business went the other way because they were waiting for the central IT
3:05
or they acquired another company and the other company was on AWS
3:09
but historically they were on Azure or vice versa or, you know
3:13
Yeah, super interesting. And what's interesting there, of course, with the larger the company, I mean, this can still be valid for a company that isn't all that large
3:22
But the larger the company, the more likely there will be some kind of a merger acquisition scenario
3:28
And you get a company that's on another cloud. And then it's kind of not really your fault
3:35
But we are kind of indicating here that maybe it isn't the best idea in the world to use multi-cloud just to use more than one cloud platform
3:46
Should we try to specify a little bit what the challenges might be if you are targeting multi-cloud
3:55
What are you facing? What's going to happen? Why is this maybe not the best idea
4:01
I think there's a couple of reasons. I mean, so my company, DevOps Group, we were AWS and Azure, you know, sort of premier partners, both
4:09
Just trying to maintain the skill sets for two clouds on the level of the people that you've got is incredibly difficult
4:18
Yes. So the skills, you know, you're doubling the amount of skills that you're having
4:24
I think the next thing is all cloud services are tiered. And so the more you consume, the lower per unit cost
4:34
So now you're spreading over too. And now I'm paying more than I would be if I had 100% on one
4:41
You know, if all of my storage was in one place, I'm going to be paying a hell of a lot less unit price
4:46
I'm going to be spending – I'm going to be having probably more bargaining power
4:52
You know, I'm going to get a better, you know, Microsoft Azure commitment, you know, whatever, EDP if you're AWS, but let's put Microsoft
4:59
Mac you know you going to have more bargaining power So I just think anybody who consciously like the number of organisations that truly need to be multi is very small
5:13
You know, even regulated financial services, you are supposed to have the ability to withstand the failure
5:23
of the outsourced relationship, which some people interpret as meaning I must be multi-cloud
5:29
It doesn't though. It's not really what it says. Yeah. It's interesting
5:33
It's absolutely, you're absolutely right with that. I couldn't, you know, agree more about the sort of the organizational issues, the woes you run into and how you lose bargaining power. That's absolutely spot on
5:48
but what about the I mean the advantages of being multi-cloud if there are such
5:56
people are trying to put arguments out there saying it's good for us to not put all the eggs in one basket
6:04
is that a valid argument? No No No It's just not a valid argument
6:15
The complexity that you are adding by trying to be multi-cloud vastly outweighs the benefits
6:24
Be multi-region, you know, so be multi-availability zone, be multi-region. If you're in a situation where, you know, you have the need for that level of availability that you need to go beyond multi-AZ and multi-region to multi-cloud
6:40
Like, if either of the or any of the big hyperscalers has failed so badly that an entire region has gone down and the region failover has gone down as well, like, that's the least of your problems
6:57
Do you know what I mean? There's going to be so much else going on. All of your dependencies and all of those other, you know, people and customers and partners that you rely on, well, they're going to be down too
7:06
So that goes away. but the complexity that you've added you know the the chances of failure goes up to the square to
7:15
the number of moving parts so the more moving parts you put in the more chances of failure
7:21
you've got so you're literally architecting in a higher chance of failure in order to avoid
7:28
a failure which is extremely unlikely to occur if you've architected your application properly with
7:33
multi multi-AZ and multi-region yeah yeah no that's that sounds that sounds reasonable and
7:39
I have another it's not a curveball but I have another fun one that I really like
7:44
in this context if you have if you manage to have a somewhat you know experienced team on
7:50
one cloud let's say we let's use your your two clouds as an example you had experience with with
7:56
using AWS you have a team on that a team that does you know or multiple teams or you have a
8:01
a center of excellence or whatnot of a cloud unit inside of your big enterprise that focuses on AWS
8:09
And then you have that one more time over again. You have another one except in Azure
8:14
So you have two of these. And you're trying to staff them. You're trying to get the right people in
8:19
And then there's this next workload or this next thing that's going to go to the cloud
8:25
And so what do you do as a company? Do you like go to both of these teams and say, hey, guys, who wants to have this thing
8:32
And I mean, what is that going to happen due to the internal politics, right
8:38
It is very rare for a customer to come to us and say, please do a technical evaluation of cloud A versus cloud B, where we're going to be best in
8:52
and the reason for that is is that the decision on which cloud you're going to go to isn't really
9:00
a technical decision anymore like if we take the example of the London Stock Exchange so the London
9:08
Stock Exchange went with Microsoft but at the same time Microsoft bought five percent or something
9:17
or whatever is it what the number was of the London Stock Exchange. You know, this was a hundreds of millions of dollar deal
9:25
on both the buying shares and also the cloud side. These deals are being done way higher above the level
9:33
of the IT director. So, you know, it's almost like, I mean, if you're a startup
9:39
sure, go to whichever one, you know, go to Azure because you're getting Azure credits through the Microsoft
9:44
for Startups program or whatever. Great. You know, it's a lot more viable to make a decision at that small scale on purely technical, whatever is the service
9:54
I prefer Azure Cognitive Services to the competitors or whatever it may be
9:59
So I think that that's, you know, but in the reality for most organisations, it's not a technical evaluation decision
10:07
They come to us to say, we've signed our Mac, you know, our enterprise agreement
10:12
We got this commitment and we now committed to spending a certain amount of money We already gone down this pathway So that not a thing as you put it I actually did maybe it not a
10:26
common thing, but I actually did come to a customer once where they indicated that they had begun to
10:34
have internal competition between the two cloud groups, which is another level of issue that you
10:41
do not want to invite into your company because that's certainly not going to, you know, amount
10:46
to anything good at all. Our cloud is better. I think having two cloud units within a single
10:55
organization is a disaster. Like if you're going to be multi-cloud or de facto that you're going
11:02
to be multi-cloud within your IT capability, they're either so far apart from each other that
11:06
it doesn't matter. Or you have one cloud unit that inherently says, you know, these are the two clouds
11:13
we support or three clouds because some people do three clouds. And then you structure it in, you know
11:19
we prefer these types of workloads to go here. Maybe we like big data in one place or ytics
11:25
in one place. We like container apps in another place. We have another, you know, another vendor
11:32
that's better for highly distributed apps, you know, with regional points of presence and things like that, you know
11:39
for low latency applications, you know, but yeah, having any kind of internal conflict, that would be bad
11:49
It would be bad. Yeah. And you shouldn't sort of fan those flames as an organization
11:55
There needs to be a strategic decision. And to that point, exactly, the current largest company
12:01
that I work with right now, they're really big. And they've made the bet that we're going with one cloud
12:09
happens to be Azure. We're going, that's why I'm there. We're going to the Azure cloud
12:14
And we will not prohibit anyone from using another cloud platform if there is a good enough reason
12:23
And that is workload dependent. For example, they have some things that would be working best
12:29
on what Oracle call a cloud. Now, I may have some issues with Oracle calling their services, their great services, their wonderful services that service so many companies a cloud
12:40
I maybe have another definition of that, but they call it a cloud service, which they offer
12:45
And I'm not going to argue that. But there are some workloads that this big company run that are running on Oracle things and would be very well suited to run on Oracle cloud
12:57
Brilliant. Do that. Now you have a reason, a real business reason, a technical reason
13:03
There is a situation. And a financial licensing reason. Probably that. Absolutely
13:10
Probably that as well. And it can be also skills related. I mean, you have people that really know this stuff really well
13:18
And you have some time to market or whatnot. There's a deadline. Something needs to be done
13:24
We know this best. Let's go there. This is what we know. This is going to work. Okay, then
13:28
do that but now you have thoroughly sat down and thought this through and you have some kind of a
13:35
reasoning and understanding but still a strategy right and and what you've said you've said the
13:41
word strategy a couple of times there yes it's strategy and leadership you know at some level
13:46
somebody has made you know I kind of talk about de facto multi-cloud versus de jure multi-cloud
13:54
Did you stumble into being multi-cloud or was it a conscious choice and you had a strategy around it
14:00
And, you know, I think most of the organizations I've seen, somebody talks about the, I know, talks about like the paved road
14:09
If you want to go with this cloud, you get a paved road and there's the repo with all the terraform to do, for patents to do everything you need to do
14:17
And there's, you know, the Azure DevOps or GitHub pipelines for everything you need to do
14:22
and it's just going to be and you inherit all the security
14:26
and all the compliance and all the monitoring or observability or whatever they call it these days, blah, blah, blah
14:31
But if you want to go over here and use DigitalOcean, well, you can
14:36
but you're supporting it yourself. You're writing all of this stuff. You're going off road
14:42
And most people at that point, they look at it and go, the cost is, well, I'm not going to go, you know
14:48
it's not that much difference. you know it's you know it's great and you bring yeah absolutely and you're bringing up such a
14:54
valid point here as well operations this is going to have to be managed and maintained by someone
14:59
so now you need to double down on those skills and those uh you know you could be a company that
15:05
actually does that uh 24 7 by yourself but i see a lot with the enterprises that they actually rent
15:11
that service they they pay someone else to to be there around the clock and so they need to
15:17
Yeah they handing that over to someone who is operating it for them And technically it going to be challenging for them to receive you know all these applications if every application they get to to to operate is
15:35
a unique snowflake it's just different each one yeah that's not going to be uh good i love how you
15:41
brought it back so we just on that we used our pricing model was a tiered pricing based on the
15:47
complexity of the application so basically it was like the you know cattle was one price pets were
15:54
another price you know you want to give us snowflakes yeah that's okay but but you're paying
15:59
for that now uh absolutely right um and that's that's the way to do it because as um as a an
16:05
operating organization if if you're if you have that business you don't want to be in that mess
16:11
when somebody just hands you this tangled spaghetti thing and like here take care of that
16:15
please. How's that going to even work? I also, I also liked, so we're coming to the end of the
16:22
episode. I'm going to let you have the last, the last say, but I really like how you brought it
16:27
back to, to leadership. When I was, I started talking about strategy, you started talking
16:32
about leadership and that's exactly what this show is about to try to untangle the messes that
16:39
come with, you know, someone saying something technical or someone's just, I don't know, I
16:44
shouldn't swear on the show but but basically just talking out of their behind because they don't know
16:50
what they're saying they're just saying something they heard they heard multi-cloud is a good thing
16:54
and multi-cloud isn't necessarily a good thing or isn't a good thing at all actually
16:59
i would say that the the the reality that you need multi-cloud certainly for a given for a single workload horses for
17:13
courses completely understand you know there's there are some technical differences and some
17:18
people you know have preferences and there's licensing reasons great you know understand
17:22
that but understand that you're buying complexity or building in complexity and everything that you
17:28
want from an operations point of view you know my background is it ops you know you want to reduce
17:34
that complexity but I think if you're a leader and and you know I've coached a number of people as a
17:42
you know, senior executives are a stakeholder, you know, go and have a look at the Azure Cloud Adoption Framework
17:48
There is such good guidance in there, you know, and just at the highest level about understanding your strategy
17:57
But when you look at CAF, like half of CAF is not related
18:04
to technology at all. It's related to organisational structure and governance and process
18:11
and all of this other stuff that you need to do. Because, you know, I think the other thing I would always say
18:20
if you're a leader and you're leading this cloud transformation is start with a blank sheet of paper
18:27
You know, don't take the same security controls. You might have the same security control objectives
18:33
but start with a blank sheet of paper. You're going to re-implement these controls in a new way
18:37
You're going to re-implement your monitoring in a new way. You're going to re-implement your networking in a new way
18:43
Whatever that thing may be, start with a blank sheet of paper
18:47
because if you try and it's cheaper if we just take this stuff
18:52
you know, from on-premise world to the cloud world. And I'm going, yeah, but, like, why are you using
18:58
some third-party network gateway that costs a fortune when you could use a default Azure one for scratch
19:05
Absolutely right. I appreciate that so much. we've had the scenario just now where they're talking about access control and how they have
19:15
central access control already set up on premises so now that we're using the cloud we're also going
19:20
to use that same system to give access to the cloud I'm like oh wow that's going to be hurtful
19:25
that's not going to work. Does that include network transit to on-premise in order to do
19:32
because I've seen I've literally seen that. Fortunately not this is just the access control
19:37
management for managing access and tracing it. But even there, it's not a good idea because you're taking data that was generated in the
19:48
cloud out of the cloud and yzing it elsewhere, which is just by no means a good idea
19:54
What you need to do is dig in and train your old security team to start to use cloud tools
20:03
for cloud problems. 100%. 100%. Brilliant. So thank you for being on the show today, Steve
20:10
It was an absolute pleasure. And I think this is going to be a gem for the leader who tries to understand
20:16
why multi-cloud isn't actually a very good thing at all. Not at all
20:22
Thank you. Thank you for having me. It was really enjoyable. Have a good evening
#Business & Industrial
#Cloud Storage
#Distributed & Cloud Computing
#Enterprise Technology
#Networking
#Web Services