Learning Library

← Back to Library

Understanding IBM Cloud Multi‑Zone Regions

Key Points

  • MZR stands for **Multi‑Zone Region**, a grouping of multiple IBM Cloud availability zones within a single geographic region.
  • An **availability zone (AZ)** is a single physical data‑center location that contains all the infrastructure required to run IBM Cloud services, including redundant fiber connectivity, power, and networking.
  • Each AZ is built with multiple telecom providers for internet ingress, dual‑source electrical feeds, generators, UPS systems for 15‑20 minute switchover, and rows of server racks with compute, storage, switches, routers, firewalls, and load balancers.
  • By distributing workloads across several AZs in an MZR, IBM Cloud provides higher availability and fault‑tolerance, ensuring services stay up even if a single zone experiences an outage.

Full Transcript

# Understanding IBM Cloud Multi‑Zone Regions **Source:** [https://www.youtube.com/watch?v=5KKoniXgHy0](https://www.youtube.com/watch?v=5KKoniXgHy0) **Duration:** 00:09:31 ## Summary - MZR stands for **Multi‑Zone Region**, a grouping of multiple IBM Cloud availability zones within a single geographic region. - An **availability zone (AZ)** is a single physical data‑center location that contains all the infrastructure required to run IBM Cloud services, including redundant fiber connectivity, power, and networking. - Each AZ is built with multiple telecom providers for internet ingress, dual‑source electrical feeds, generators, UPS systems for 15‑20 minute switchover, and rows of server racks with compute, storage, switches, routers, firewalls, and load balancers. - By distributing workloads across several AZs in an MZR, IBM Cloud provides higher availability and fault‑tolerance, ensuring services stay up even if a single zone experiences an outage. ## Sections - [00:00:00](https://www.youtube.com/watch?v=5KKoniXgHy0&t=0s) **Explaining IBM Cloud Multi‑Zone Regions** - A product manager describes what an MZR (Multi‑Zone Region) is by defining availability zones and outlining the physical data‑center infrastructure, fiber connectivity, and power resources that support IBM Cloud services. - [00:03:08](https://www.youtube.com/watch?v=5KKoniXgHy0&t=188s) **Cooling Essentials for Multi‑Zone Regions** - The passage stresses that efficient air conditioning is crucial for data‑center operation and defines a Multi‑Zone Region (MZR) as a single geography—like Dallas‑Fort Worth, Frankfurt, or London—containing three or more physically separated, fault‑isolated Availability Zones. - [00:06:15](https://www.youtube.com/watch?v=5KKoniXgHy0&t=375s) **Why Multi-Zone Architecture Matters** - The speaker explains that deploying cloud‑native workloads across multiple availability zones ensures high availability and resilience against failures such as power outages or natural disasters, using load balancers to automatically route traffic when a zone goes offline. - [00:09:30](https://www.youtube.com/watch?v=5KKoniXgHy0&t=570s) **Polite Closing Gratitude** - The speaker expresses continued appreciation, indicating gratitude at the end of an interaction. ## Full Transcript
0:00Hi there, and thanks so much for watching our video. 0:02My name is Bradley Knapp and I am one of the product managers here at IBM Cloud. 0:07And the question that I'm answering for you today is what is an MZR? 0:11You're obviously on our site and you're seeing this mentioned a lot. 0:13So what is it? 0:16M. 0:17Z. 0:17R. 0:19Well, MZR is an acronym, it stands for Multi-Zone Region. 0:23And so before I can answer, "what is a multi-zone region?", 0:27first, we need to talk about "what is a zone?", 0:30or specific to IBM Cloud, "what is in availability zone?", 0:32or an AZ. 0:35So, an availability zone is a single physical location 0:40that contains all of the infrastructure necessary to run 0:43IBM's Cloud. 0:46So if we think about what goes inside an AZ, 0:48let's imagine our building here, right? 0:51So this is a physical data center building. 0:54Well, what do you need to run the cloud? 0:55Obviously, you've got to have some internet that comes in, right? 0:58So you've got fiber that's coming in from multiple providers, 1:02multiple redundant 10 gigabit links. 1:04And so we'll have fiber that come in 1:06from all of the major telcos 1:08because we need to guarantee as much uptime as possible. 1:12Also, because you, our clients need to be able to tie in with us, 1:15you want to come in across our private network. 1:18And so we want to be able to support the telco of your choice as you tie in. 1:22So we've got a building, we've got some fiber running to it. 1:25What else do we need? 1:26We need electricity, right? 1:28And we don't want just a little bit of electricity. 1:31We want a lot of electricity and we want it redundant. 1:34And so we'll have electricity that comes into this zone 1:36from two different directions. 1:38We'll have one from Provider A and one from Provider B. 1:42And then in case we have problems with that electricity, 1:45we've also got to have generators. 1:49Because if the power goes out, 1:50we can't just shut the data center down, 1:52we still have to generate electricity, 1:54so we're going to have generators outside the building. 1:56We have to have enough generating capacity to run everything that's inside. 2:00And in addition to those generators, 2:02in order to bridge the gap between when the power goes off 2:05and the generators go on, 2:07we have huge UPS systems. 2:09UPS is an Uninterruptible Power Supply, 2:12and those UPS systems are sized 2:15to get us 15 to 20 minutes between when the power comes off 2:18and the generators come all the way online. 2:20Now, in practice, we don't need anywhere near that much time. 2:23But it's just a good thing to have. 2:25So we've got our power. 2:27We've got our internet. 2:28What comes next? 2:29Well, we need the compute and the storage, right? 2:32We've got the network, well, the outside part of it. 2:35We need the compute and we need the storage. 2:37So inside this data center, we have rows of servers. 2:42And in these rows are racks, 2:45and in each rack are up to 43 servers. 2:49In addition to the servers that are on those rows, 2:52we also have all of the network gear: 2:55the switches, the routers, 2:56the hardware firewall devices, hardware load balancers. 3:00So all of the equipment, 3:02what you think of is the magic that is the underlying layer of the cloud, 3:06that is all in these rows. 3:08And then the last piece that we have is arguably the most important, 3:13specifically depending on region, and that's going to be - 3:16I'm going to put that over here - that's our air conditioning. 3:19Because what's the point of having a data center 3:21with all of this compute, storage and network capacity in it 3:24if you can't keep it cool? 3:26So we've got air conditioning over here, keeping everything cool. 3:29It's blowing cold air into the cold aisles, 3:31we're ducting the hot air out of the hot aisles to get rid of it. 3:34And so this is an Availability Zone. 3:38This is one AZ. 3:40So, to get back to our original question, 3:43"what is an MZR?" 3:45Well, an MZR, a Multi-Zone Region, is a single geography. 3:50So, a single area like Dallas-Fort Worth, 3:53or Frankfurt, or London, 3:56where we will have three or more availability zones. 4:00And so we're going to have Availability Zone 1, 4:04Availability Zone 2, 4:06and Availability Zone 3. 4:08One, two, three. 4:11Again, each of these is physically distinct from the others, 4:14and we try and put some geographical separation in them. 4:17Anywhere from, oh, 5 or so kilometers 4:20up to about 25 kilometers distance between them. 4:24And so, not only do you have physical distance, 4:26but we also make sure that they are in geographically separate fault profiles. 4:31So for an area that is active seismically, 4:34we want to be sure that they don't share common seismic faults. 4:37If we're in an area that has flooding risks, 4:39we want to be sure that they're not all sitting in the same flood plain. 4:42If we're in a politically unstable area, 4:44we want to be sure that they each have their own security posture. 4:47So we've got our three zones. 4:50Now we have to interconnect them all. 4:53Yeah, let's use a different color to interconnect. 4:55We're going to interconnect them like this. 4:58So we have all of this interconnection. 5:00This guarantees that if you've got a workload 5:03that is running in zones 1 and 3, you can keep them in sync. 5:07But we're not done. 5:08I talked earlier about customers that want to come in 5:11and they want to bring their fiber to us. 5:14They don't just want to come in over a VPN, 5:16they want a dedicated MPLS circuit or something like that. 5:19So in addition to our three zones, 5:22we are also going to have 2 network points of presence 5:27and I'm going to call these A and B. 5:30A "Network Point of Presence" or network PoP, 5:33or a network hotel, 5:34is a place where a lot of fiber comes together. 5:37Fiber from companies that lease it individually, 5:40fiber from the telcos, fibers from the provider. 5:43But the fiber comes in here. 5:45These are the main connection points. 5:47Now, in some IBM Cloud MZR locations, 5:50we're going to co-locate our network hotel in with the data center, 5:53and in other ones they're disparate, 5:55but each of these locations is also going to be fully interconnected 6:01with the data centers because again, 6:03we have to have as much redundancy as possible. 6:07We don't want any single problem to take out a workload. 6:10So this is the construct, the "multi-zone region". 6:15Now the last question is, 6:18"why would you care about a multi-zone region?" 6:21Why does this matter? 6:22And so this is where we're going to get into the specifics 6:25of running highly available cloud native workloads. 6:30Because this is a different construct 6:31than how you did it on-premises back in the day. 6:34So a cloud workload that is run highly available. 6:38Let's get our data centers back up here again. 6:40So we've got 1, 2, and 3. 6:46In a cloud native architecture, 6:48you never run in just a single zone unless you have to, 6:52and sometimes there are good reasons that you have to. 6:54Some IBM Cloud locations only have one zone, 6:57that's called a single zone region, 6:59and there are lots of good reasons for that. 7:01But we're going to focus on the multi-zone architecture. 7:03So let's say that your workload, 7:06your primary is going to be Zone 1, right? 7:10So if you've got your little guy over here, 7:14and he's surfing your website on his laptop 7:18and typing away and all that, his actual request, 7:21the request transmission themselves 7:23are going to be coming in over the public internet into Zone 1. 7:31Now it's going to get processed here, 7:34data is going to get passed back. 7:36What happens if something goes wrong? 7:38That could be a power outage. 7:40That could be some sort of a natural disaster, 7:42a flood, a hurricane. 7:44It could be anything that goes wrong 7:46that could cause Data Center 1, 7:49or Availability Zone 1, to come offline. 7:52Well, if your architecture is correct, 7:55you aren't using just one. 7:57You have a load balancer involved. 7:58You're going to use one and two, 7:59or one and three, or one, two and three. 8:02So instead of going directly to one, 8:04let's say that that connection then fails 8:07because you've got some redundancy 8:08and it says, "Oh, this one fell offline. 8:10Now, instead of going to 1, no more, now we're going to go to 2". 8:15And that's going to be the location of where we're going to do the compute, 8:19the app that's serving up the data for the app that you've written, 8:21whatever it may be. 8:23Likewise, if there's a problem with 2, it goes offline. 8:26That's OK because you've already got your redundancy. 8:28So it's going to start going to 3. 8:30This cloud native, highly available architecture 8:33is key to a successful cloud deployment. 8:36If you are not running your production apps 8:39highly available in two zones, 8:41you are going to get yourself into trouble. 8:43Now, if you have additional questions about how to engage 8:46in this kind of architecture, we would love to help you out. 8:49Please reach out to your IBM sales team. 8:51Go check out the IBM Cloud Architecture Center. 8:54We have lots of documentation 8:55on how to build your workload 8:57so that you don't have to worry about a failure in any particular zone. 9:01You are set up for native success in a cloud native manner. 9:07Thank you so much for your time today. 9:09If you have any questions, please drop us a line below. 9:13If you want to see more videos like this in the future, please do like and subscribe and let us know. 9:18And don't forget: you can grow your skills and earn a badge with IBM Cloud Labs, 9:23which are free, browser-based interactive Kubernetes labs, 9:27that you can find more information on by looking below. 9:30Thanks again!