Azure real world questions

[DA-LION-619]

This is ping form my house.
So what would be so sensitive that it needs less than ~3ms latency

Exactly, it's a ping from your house, it's dependent on physics.
It says nothing about variation(jitter) or the route taken, try your test from a non-peering network http://lg.he.net/
You can execute SQL and create a response within ~3ms.

You can't get into the site unless you have been authorised.
Opening the doors sets of alarms that notify people on site and remotely (in other countries, so it isn't something that goes unnoticed)

Huh?

From the above description there are aspects to why some things have to be on-prem,
Designed to work in a LAN, internet connectivity not required
Use of hardware, like hardware security modules etc.

To have been authorised, you first would need to be verified(1:1) or identified(1:N). Whatever hardware is in use(keypads, fingerprint scanners), it would be communicating over the network, typically TCP sockets.
Authorisation is a JIT event, getting in after-hours would be an exception similar to setting off alarms.
Notifications have to be acknowledged and those on-site are a priority.

^That doesn't seem like a CPU bound workload that would benefit from the cloud. It's I/O bound.
Sure it could be 3ms, but that's 3ms of back pressure and you can't drop security like packets.

 

[ubercal]

Exactly, it's a ping from your house, it's dependent on physics.
It says nothing about variation(jitter) or the route taken, try your test from a non-peering network http://lg.he.net/
You can execute SQL and create a response within ~3ms.



From the above description there are aspects to why some things have to be on-prem,
Designed to work in a LAN, internet connectivity not required
Use of hardware, like hardware security modules etc.

To have been authorised, you first would need to be verified(1:1) or identified(1:N). Whatever hardware is in use(keypads, fingerprint scanners), it would be communicating over the network, typically TCP sockets.
Authorisation is a JIT event, getting in after-hours would be an exception similar to setting off alarms.
Notifications have to be acknowledged and those on-site are a priority.

^That doesn't seem like a CPU bound workload that would benefit from the cloud. It's I/O bound.
Sure it could be 3ms, but that's 3ms of back pressure and you can't drop security like packets.

eg: plenty companies use software like pastel evolution.Its very sensitive to latency and jitter so accessing across a vpn is a no no.This is the reason why hybrid is the best as it gives you flexibility depending on the specific workload you dealing with.People that say the cloud is the answer to all your problems are just trying to sell you a pipe dream , doesnt work like that in the real world.

 

[Barbarian Conan]

My guess is AWS and Azure are pretty close on pricing so doubt that is what he means.

A lot of "old school"/"IT guys" believe that they are being ripped off by AWS or Azure.

I'm on the AWS side as you, know little to nothing about Azure or their ecosystem, so I can only comment from the AWS perspective.

But you can't put a price on not needing to worry about your "dedicated server" having a hardware failure, then you need to email the company because they don't have monitoring. Then their connections go down. No virtual networking, etc. etc. etc.

Comparing these local dedicated hosting options to AWS is like comparing a bicycle to a top of the line car.
They aren't in the same league, not even in terms of use cases.

I also put dedicated in quotes because some of them have fine print where you actually get part of the machine. Don't even want to imagine what the security, availability, reliability & consistency is like in that situation.

Part of the value proposition of AWS and Azure is that you need fewer "IT Guys" that you have to pay.

 

[DA-LION-619]

Part of the value proposition of AWS and Azure is that you need fewer "IT Guys" that you have to pay.

 

[ubercal]

Part of the value proposition of AWS and Azure is that you need fewer "IT Guys" that you have to pay.

you are 100% wrong.It just means things are virtualized.You still need core skills like networking , security etc.Just that its now in a virtualized environment and you can sit in the comfort of your chair and design an entire virtual network without lifting a finger.

 

[ToxicBunny]

eg: plenty companies use software like pastel evolution.Its very sensitive to latency and jitter so accessing across a vpn is a no no.This is the reason why hybrid is the best as it gives you flexibility depending on the specific workload you dealing with.People that say the cloud is the answer to all your problems are just trying to sell you a pipe dream , doesnt work like that in the real world.

Beyond just things like latency and jitter that make hybrid the best approach, there are data sovereignty issues and regulations in certain areas that need to be adhered to, and also in many ways just base cost. Cloud will always be more expensive for permanently running systems that don't "breathe" in terms of scale.

 

[Barbarian Conan]

you are 100% wrong.It just means things are virtualized.You still need core skills like networking , security etc.Just that its now in a virtualized environment and you can sit in the comfort of your chair and design an entire virtual network without lifting a finger.

Not me, it's part of what MS sells.
But in theory it is correct. You don't need someone handle any of the physical stuff.
For every step you take in IaaS, PaaS, and eventually SaaS, there is less that your own staff have to do.

 

[Gnome]

you are 100% wrong.It just means things are virtualized.You still need core skills like networking , security etc.Just that its now in a virtualized environment and you can sit in the comfort of your chair and design an entire virtual network without lifting a finger.

Having a data centre and/or on premises servers requires a lot of extra people you no longer need.

 

[Lord Flacko]

I’m an advocate for Cloud, I don’t see the need for on-prem infrastructure to be honest. Running workloads / services in the cloud is the way to go. There’s a shift from capex to opex. No more hardware to maintain, if you’re crazy about virtual machines you can use auto scaling to meet performance requirements. There’s plenty of benefits to using the cloud, just don’t get caught up with staying in the past (remember Blockbuster LLC & Kodak).

 

[ubercal]

I’m an advocate for Cloud, I don’t see the need for on-prem infrastructure to be honest. Running workloads / services in the cloud is the way to go. There’s a shift from capex to opex. No more hardware to maintain, if you’re crazy about virtual machines you can use auto scaling to meet performance requirements. There’s plenty of benefits to using the cloud, just don’t get caught up with staying in the past (remember Blockbuster LLC & Kodak).

youre a developer right ?

 

[JohnStarr]

Part of the value proposition of AWS and Azure is that you need fewer "IT Guys" that you have to pay.

Yes, and no. You still need monitoring of what you have in Azure. Still need patching. Still need deployment. But it does give you the opportunity to have your IT team learn new skills. Automation being one of them (which to choose, is another story). You can always retread someone in the IT space for the better.
If they choose not too, then they will eventually be caught out with a lack of skills.

 

[PsyWulf]

I think the "fewer" is hardware guys
Nobody to rack-n-stack or go change harddrives,or pull in network/FO cables

 

[JohnStarr]

I’m an advocate for Cloud, I don’t see the need for on-prem infrastructure to be honest. Running workloads / services in the cloud is the way to go. There’s a shift from capex to opex. No more hardware to maintain, if you’re crazy about virtual machines you can use auto scaling to meet performance requirements. There’s plenty of benefits to using the cloud, just don’t get caught up with staying in the past (remember Blockbuster LLC & Kodak).

Yes, and no. You will always need some form of hardware on the site, be it for authentication via AD (what happens if your VPN to your cloud platform goes down?), or large-scale ERP systems.
Hence hybrid. And this is advocated by many companies. Lift-and-shift all your workloads just isn't economically viable for many, and this includes very large-scale companies.

 

[Lord Flacko]

youre a developer right ?

I wouldn’t say so. Calling myself a developer is a stretch. I’m just anti IAAS, the solutions I design and develop mostly utilise serverless technologies (along the lines of AWS Lambda & Azure Functions), I focus on writing code and let the cloud provider provision and maintain the underlying stack that the code I deploy runs on.

Yes, and no. You will always need some form of hardware on the site, be it for authentication via AD (what happens if your VPN to your cloud platform goes down?), or large-scale ERP systems.
Hence hybrid. And this is advocated by many companies. Lift-and-shift all your workloads just isn't economically viable for many, and this includes very large-scale companies.

Lift and shift doesn’t utilise the true capabilities of the cloud, applications need to be rearchitected to run in the cloud. I’m sure SAP also runs fine in the cloud, as for VPN’s and the like make sure the setup is a highly available configuration, cutting costs and deploying solutions that have a single point of failure leads to disappointment.

 

[JohnStarr]

I wouldn’t say so. Calling myself a developer is a stretch. I’m just anti IAAS, the solutions I design and develop mostly utilise serverless technologies (along the lines of AWS Lambda & Azure Functions), I focus on writing code and let the cloud provider provision and maintain the underlying stack that the code I deploy runs on.


Lift and shift doesn’t utilise the true capabilities of the cloud, applications need to be rearchitected to run in the cloud. I’m sure SAP also runs fine in the cloud, as for VPN’s and the like make sure the setup is a highly available configuration, cutting costs and deploying solutions that have a single point of failure leads to disappointment.

Well, SAP runs just fine in the cloud based on experience. But the infrastructure to do so for a large company in terms of redundant lines etc. can become very expensive. Cloud does not always equate to cost savings.
It is, however, the way to go. Either hybrid, or for a small environment, the whole hog.

 

[ToxicBunny]

I’m an advocate for Cloud, I don’t see the need for on-prem infrastructure to be honest. Running workloads / services in the cloud is the way to go. There’s a shift from capex to opex. No more hardware to maintain, if you’re crazy about virtual machines you can use auto scaling to meet performance requirements. There’s plenty of benefits to using the cloud, just don’t get caught up with staying in the past (remember Blockbuster LLC & Kodak).

In all honesty, Cloud brings lots of benefits to the table but it has caveats for many companies and situations.

And yes, at a basic level it moves from Capex to Opex, but even that isn't as cut and dried as it seems. If you want to take the best advantage of the Cloud and get the best pricing you move back into a hybrid Capex/Opex world with reserving base capacity for a defined period from your provider.

There are also some significant compliance and regulatory things that a Cloud provider can't necessarily ever solve for you which is where some level of On-Prem equipment will be needed for certain types of companies and markets.

 

[DA-LION-619]

Facebook has entered the chat

 

[The_Ogre]

@TribbleZA Did you re-write?

Not sure if she every replied to you in private, but she was in hospital (COVID) for quite some time, so doubt she rewrote.

FWIW, she's been out for a while and back to her busy self

 

[randomcat]

Not sure if she every replied to you in private, but she was in hospital (COVID) for quite some time, so doubt she rewrote.

FWIW, she's been out for a while and back to her busy self

She didn't reply. Shoot, that is hectic.

 

[Gnome]

And yes, at a basic level it moves from Capex to Opex, but even that isn't as cut and dried as it seems. If you want to take the best advantage of the Cloud and get the best pricing you move back into a hybrid Capex/Opex world with reserving base capacity for a defined period from your provider.

I don't see how that is possible. Backup power generation, HVAC, physical security, replacing hardware (end of life), power costs, staff that maintain the servers on-site (above and beyond the guys that access them remotely as you would in the cloud).

In SA I can't see how any company can claim given the above that it is cheaper to have your own cloud. The scale just isn't there.

There are also some significant compliance and regulatory things that a Cloud provider can't necessarily ever solve for you which is where some level of On-Prem equipment will be needed for certain types of companies and markets.

Compliance and regulation is much cheaper in cloud.
Amazon has regions with special compliance just for that. ie. ITAR compliance data centers.
At their scale they can afford to build the expensive software tooling to make the data center completely hands off and automated in every way.
From rotating credentials, delivering hands-off credentials, monitoring for bad servers, automated scrubbing of machines when you are done with them, the list is so long I could write pages about it.

Each one of those things has a software cost associated with it. If you had to pay that cost as a company, you'd ultimately be spending billions, which is what Amazon is spending. And when that stuff isn't bringing you in money, it is impossible to justify. Whereas for Amazon it is super easy to justify. That is ultimately why hybrid or on-prem is false economy. Amazon backed down from their position that on-prem doesn't make sense because too many customers just couldn't accept that fact. But it is the less secure option, there is no 2 ways about it. On-prem was how Oracle and M$ tried to sell cloud before they had any meaningful way to compete with Amazon. It didn't hurt that a bunch of "IT techs" were just looking for a reason to say "see I told you so". I can see very few situations in which on-prem could make sense