Learning Library

← Back to Library

IaaS Explained: Compute, Storage, Network

Key Points

  • IaaS (Infrastructure as a Service) lets you rent the core building blocks of cloud—compute, storage, and networking—rather than buying and maintaining physical hardware.
  • The “as‑a‑service” part describes the on‑demand, usage‑based billing model, similar to other offerings like PaaS (Platform) and SaaS (Software).
  • Compute resources in IaaS are categorized into general‑purpose servers, GPU‑accelerated instances for AI/ML workloads, and high‑performance computing (HPC) instances for tasks requiring very high clock speeds and core counts.
  • Storage is offered in various tiers (e.g., block, file, object) to match differing performance and durability requirements, while networking provides the essential connectivity that links compute and storage together.

Full Transcript

# IaaS Explained: Compute, Storage, Network **Source:** [https://www.youtube.com/watch?v=XRdmfo4M_YA](https://www.youtube.com/watch?v=XRdmfo4M_YA) **Duration:** 00:08:07 ## Summary - IaaS (Infrastructure as a Service) lets you rent the core building blocks of cloud—compute, storage, and networking—rather than buying and maintaining physical hardware. - The “as‑a‑service” part describes the on‑demand, usage‑based billing model, similar to other offerings like PaaS (Platform) and SaaS (Software). - Compute resources in IaaS are categorized into general‑purpose servers, GPU‑accelerated instances for AI/ML workloads, and high‑performance computing (HPC) instances for tasks requiring very high clock speeds and core counts. - Storage is offered in various tiers (e.g., block, file, object) to match differing performance and durability requirements, while networking provides the essential connectivity that links compute and storage together. ## Sections - [00:00:00](https://www.youtube.com/watch?v=XRdmfo4M_YA&t=0s) **Understanding IaaS: Infrastructure Basics** - Bradley Knapp explains what IaaS stands for, how it differs from other cloud service models, and outlines its core infrastructure categories such as compute. - [00:04:48](https://www.youtube.com/watch?v=XRdmfo4M_YA&t=288s) **Object Storage and Multi‑Tenant AI Service** - The speaker describes streaming a billion images from inexpensive object storage to GPU servers for model training and back, then storing results, and explains that this capability is offered as a shared, multi‑tenant “as‑a‑Service” solution. ## Full Transcript
0:00All right, hi everybody and welcome back to the channel. 0:02My name is Bradley Knapp, and I'm one of the Product Managers here at IBM Cloud, 0:05and what I want to talk with you guys about today 0:07is a question that we get fairly commonly 0:09when folks are starting their cloud journey 0:11and starting to learn about cloud, 0:12and that's, "What is IaaS?" 0:15I read about cloud, I see this IaaS thing everywhere, 0:18what does it actually mean? 0:20And, so, "IaaS" is an acronym, and so it's broken into 2 parts: 0:23the first part, the "I", that's "infrastructure". 0:26And, so, if you think of cloud 0:28as being just some other dude's computer, 0:30running somewhere else, 0:31that's the infrastructure part. 0:33And, so, that infrastructure, if it's not cloud - 0:36it could be running in a data center somewhere, 0:39it can be running in a closet somewhere, 0:41your laptop or your desktop is infrastructure. 0:44And then the "a.a.S." piece is "as-a-Service". 0:47That's the billing method, that's the way that you consume it. 0:51And there are other kinds of as a service. 0:54You've got "PaaS", Platform as a Service, 0:56you have "SaaS", Software as a Service. 0:58There's lots of different kinds of things that you can consume as-a-Service 1:03but very specifically what we want to talk about 1:05is the "I", it's the infrastructure. 1:07And so I've got this diagram written out over here 1:10because infrastructure really falls into three 1:13main categories, right. 1:15The first category is going to be compute, that's where the 1:18processors are that's where the actual lifting and computing gets done. The 1:23second piece which is storage kind of falls into three main buckets and lots 1:27of smaller buckets on top of it because there's different kinds of storage. And 1:30then the third piece, the piece that ties everything together 1:34that's our network piece. And so we're gonna draw this one over here because 1:38without network you can't do anything, network is how the compute talks to the 1:42storage and that's how the compute talks to the other compute. And so like I said 1:47we can break this down into different pieces and so in the 1:51compute side I've got three things called out up here, the first one is kind 1:55of I've just got a labeled compute, its general purpose compute, right. This is 1:59your normal web server or application server, it can really be whatever general 2:03purpose kind of computing needs you have. The second two are, or the second and the 2:08third really are more specific right. So GPU is 2:12is a graphics processor, that's a very very high speed processor that's used in 2:17conjunction with a traditional processor for specific kinds of workloads right. 2:21This is gonna be your machine learning and your AI. And then the third piece, HPC, 2:26that's high-performance computing. So there's specific kinds of workloads that 2:30had very specific requirements as far as frequency which is your clock speed and 2:34the number of cores that are required where you have to have lots of power 2:38packed into a very, very, very small footprint, that's gonna be your HPC. And 2:41likewise on the storage side you've got different kinds of storage because you 2:46have different storage needs. The most commonly used one is gonna be object 2:50storage. Object storage is a little bit lower performance but it's relatively 2:55inexpensive and that's for your general purpose storage right. What goes into 2:59object storage? Well you can have things like pictures, you can have documents, you 3:05can have really whatever you want can go into that object storage, it's where all 3:09of the data and all of the graphics on the web server that's all hiding 3:14in object storage. And then the second and the third piece that I've got called 3:19out here, block and file, these are specific kinds of storage, specific kinds 3:24of network storage, and they attach in very specific ways. Block storage 3:28attaches with iSCSI, file storage attaches with NFS, it's the way that they 3:32mount into the actual compute itself. And there are specific kinds of applications 3:37that require block storage or file storage because each of them has their 3:41own features and benefits. And so to talk about how we pull all of these things 3:46together we need to talk about the network, because network has two main 3:52components that matter. And so what I want you to do is I want you to think of 3:55your network as a pipe, right. And so a network can be a small pipe, that would 4:00be like a pipe measured in megabits so you can't press much data through it. Or 4:05it can be a very large pipe, that very large pipe that would be measured in 4:08gigabits per second. And so the more data you need to push simultaneously the 4:14larger pipe you need and the more bandwidth you need. The second way that 4:18we measure network traffic is how much data gets pushed through this pipe over 4:23a set period of time. Normally it's billed by the 4:26month but it could also be billed by the minute, by the second, or maybe even by 4:30the day, or by the week. And so to take all of this and tie all this together, I 4:34want to use an example of something that requires some specialty components right, 4:39we're going to talk a little bit about an AI workload. And so if you think about 4:43an AI workload where you're going to do automatic visual recognition of pictures. 4:48Let's say that you have a billion pictures down here in object storage 4:52that you are then going to use to train your model that's running on these GPU 4:57servers, and so you take that billion pictures and since a billion is a lot 5:01and pictures are very large you have to push them through a really big pipe, that's 5:05your network pipe up into the GPU server but the GPU server doesn't have any 5:11storage inherent to it. So that GPU server is actually going to take and 5:15write that into block. And it's going to write that data back and forth, and back 5:21and forth until the model is done. Once it's trained it's going to take all of 5:26the data that we pushed up here and all of the results and it's going to write 5:30all of that back down into object storage. Why object storage? Because 5:34again, it's less expensive, it's a good archiving solution. And so you're pushing 5:39a ton of data through these pipes while they're turned on and then once you're 5:43done you get rid of them. And so the second piece that I want to talk about 5:48is the "as-a-Service" piece, this is the way that you consume. And so when we talk 5:54about as-a-Service there are kind of four things that really, really matter in 5:57this model and the first one is that offerings that are consumed as-a-Service 6:03are generally speaking shared. And so by shared I mean they're multi-tenant, many 6:08people use the same offering, we just take and carve it up and make it 6:12available to multiple different customers simultaneously. So that's the 6:15first piece of as-a-Service. The second piece is the hourly or monthly piece. 6:21This is talking about how we bill. In the case of compute, it could be a 6:26certain number of cents, or certain of dollars per hour, or per month. In the 6:30case of storage, we would bill out in the amount of data that's stored in a given 6:34month, so that would be cents per gigabyte per month. In the case of 6:38network, there are two different metrics we 6:40about earlier right. The size of the pipe you would pay per month charge for that, 6:44and then the amount of data that goes through it again measured in gigabytes 6:48per month, or cents per gigabytes per month. So that's our billing metric. And 6:53then the third piece, and this is a very important one, is that there are no 6:56contracts involved in an as-a-Service model, or there aren't necessarily 7:01contracts. There can certainly be them but they're generally advantageous to 7:05you. By no contracts we mean that you don't have to agree to use something for 7:10a set amount of time, you use it for as long as you need it and then you get rid 7:14of it. And so rather than a checkmark for no contracts I'm just gonna put a little 7:18X there. You only use it to when you need it, it's on demand. And then the last 7:23piece, and this is probably the most important as-a-Service offerings are 7:28self service. That means that you can go out to a website, you punch in your 7:32information, your payment details, click the Go button and that as-a-Service 7:37offering is going to be provisioned and delivered to you. It's not something that 7:41takes days or weeks or months to set up and configure, it's one that can be 7:45provided in minutes or hours. Thanks for stopping by today. If you have any 7:49questions, please drop us a line below in the comments. If you want to see more 7:53videos like this in the future, please do hit that like and subscribe button that 7:57way we know you're interested in it. And don't forget you can always get started 8:00and cloud at no cost by signing up for a free IBM Cloud account at cloud.ibm.com.