Best of 2023: OCI Compute and Load Balancing

In this episode, Lois Houston and Nikita Abraham, along with Rohit Rahi, look at two important services that Oracle Cloud Infrastructure provides: Compute and Load Balancing. They also discuss the basics of instances.

 

Oracle MyLearn: https://mylearn.oracle.com/

Oracle University Learning Community: https://education.oracle.com/ou-community

X (formerly Twitter): https://twitter.com/Oracle_Edu

LinkedIn: https://www.linkedin.com/showcase/oracle-university/

 

Special thanks to Arijit Ghosh, Kiran BR, David Wright, the OU Podcast Team, and the OU Studio Team for helping us create this episode.

 

--------------------------------------------------------

 

Episode Transcript:

00:00
Welcome to the Oracle University Podcast, the first stop on your cloud journey. During this series of informative podcasts, we’ll bring you foundational training on the most popular Oracle technologies. Let’s get started.
00:26
Nikita: Hello and welcome to the Oracle University Podcast. I’m Nikita Abraham, Principal Technical Editor with Oracle University, and with me is Lois Houston, Director of Innovation Programs.
Lois: Hi there. You’re listening to our Best of 2023 series, where over the last few weeks, we’ve been revisiting our most popular episodes of the year.
00:47
Nikita: In today’s episode, which is #5 of 6, we’ll listen in to a conversation Lois and I had earlier this year with Rohit Rahi, Vice President of CSS OU Cloud Delivery, on OCI Compute and Load Balancing. We began by asking Rohit why one would use Load Balancer.
Lois: So let’s get right to it!
01:06
Rohit: You would use Load Balancer to achieve high availability and also achieve scalability. 
So typically the way Load Balancer works is, they're also referred to as Reverse Proxies, you would have a Load Balancer, which would be used accessed by multiple clients, various clients. And these clients would hit the Load Balancer, and the Load Balancer would proxy that traffic to the various backend servers. So in this way, it not only protects the various backend servers, but also provides high availability. In case a particular backend server is not available, the application can still be up and running. And then it also provides scalability because if lots of clients start hitting the Load Balancer, you could easily add more backend servers. And there are several other advanced capabilities like SSL termination and SSL passthrough and a lot of other advanced features. 
So the first type of Load Balancer we have in OCI is a layer 7 Load Balancer. Layer 7 basically means it understands HTTP and HTTPS. That's the OSI model. And then there are various capabilities available here. 
02:13
Nikita: The Load Balancer comes in two different shapes, right? Can you tell us a little about that?
Rohit: One is called a flexible shape where you define the minimum and the maximum and you define the range. And your Load Balancer can achieve any kind of-- support any kind of traffic in that particular range, going from 10 Mbps all the way to 8 Gbps. 
The second kind of shape is called dynamic where you predefine the shapes. So you have micro, small, medium, large, and you predefine the shape. And you don't have to warm up your Load Balancer. If the traffic comes to that particular shape, the Load Balancer automatically scales. 
02:53
Rohit: You can always do a public and a private Load Balancer. Public means Load Balancer is available on the web. Private means your multiple tiers, like a web tier, can talk to your database tier and balance the traffic between them, but both tiers don't have to be public. 
A Load Balancer is highly available, highly scalable by design.
03:12
Lois: And what about the second type of Load Balancer?
Rohit: The second kind of Load Balancer we have in OCI is called the Network Load Balancer. And as the name specify, Network Load Balancer operates at layer 4, layer 3, and layer 4 so it understands TCP, UDP, also supports ICMP. Again, like HTTP Load Balancer, it has both public and a private option, so you could create a public Network Load Balancer or a private Network Load Balancer. It's highly available, highly scalable, all those features are supported. 
03:42
Nikita: Now, why would you use Network Load Balancer over an HTTP Load Balancer? 
Rohit: The primary reason you would use it is it's much faster than HTTP Load Balancer. It has much lower latency. So if performance is a key criteria for you, go with Network Load Balancer. 
On the contrary, the HTTP Load Balancer has higher level intelligence because it can look at the packets, it can inspect the packets, and it gets that intelligence. So if you're looking for that kind of routing intelligence, then go with HTTP Load Balancer. 
04:15
Rohit: So OCI Compute service provides you virtual machines and bare metal servers to meet your compute and application requirements. The three defining characteristics of this service include this scalability, high performance, and lower pricing.
So the first thing in the OCI Compute service is you have this notion of flexible shape. What does it mean? Well, it means you could choose your own course, your CPU processors, and you could also choose your own memory. Literally, there are thousands and thousands of configurations you can choose from.
04:49
Lois: But what’s the use of doing this? 
Rohit: The use of doing this is you could select the right machine type by using our flexible shapes. 
And in the cloud, there's this notion of T-shirt sizing. So you have a small, medium, large kind of shapes, and your application has to fit those shapes. And sometimes you overprovision or underprovision, and you have to go through that painful process of changing your machine types. We hope with this flexible shapes, you don't have to do that. 
05:20
Rohit: If you still want to use the traditional approach, we have virtual machines, we have bare metal servers, and we have dedicated host. And you could use either one of them or all of them. And bare metal servers basically means you get a full machine, a full server which is completely dedicated to you. Dedicated host basically means that you get a full dedicated bare metal machine. But on top of that, you could run virtual machines. 
Not only this, but OCI is only one of the two cloud providers to provide you options on processors. So you can run AMD-based instances, you could run Intel-based instances, and you could also run Arm-based instances-- are really a powerful thing for mobile computing. The phones you are using today are probably running on Arm processors. Now, Arm is coming into the data centers.
06:16
Have something interesting to share with the Oracle University Learning Community? Present your topic at an exclusive community event. Help yourself by helping others. Start building your reputation and personal brand today. 
If you are already an Oracle MyLearn user, go to MyLearn to join the community. You will need to log in first. If you have not yet accessed Oracle MyLearn, visit mylearn.oracle.com and create an account to get started.
06:48
Nikita: What can you tell us about the pricing of this, Rohit?
Rohit: On the pricing side, the service implements pay-as-you-go pricing. We are 50% cheaper than any other cloud out there, just to begin with. And not only that, you could use something like a Preemptable VMs to reduce your cost by more than 50% from your regular instances. 
Preemptable VMs are low cost, short lived VMs suited for batch jobs and fault tolerant workloads. These are similar to regular instances, but priced 50% lower. So you can use them to reduce your cost further.
So when we say an instance, what we mean is a compute host. And it has several dependencies. So let's look at them. 
07:31
Rohit: So you have an Oracle Cloud region here. A region is comprised of multiple ADs. An AD is nothing but a data center.
The first dependency the compute service has or compute hosts have is on Virtual Cloud Network. So in order to spin up a compute instance, you need a Virtual Cloud Network.
You have a network divided into smaller portions called subnets. So you have a subnetwork here, and you need to create these before you can spin up a compute host. 
08:00
Rohit: Now you can spin up a compute host. It's a physical construct. Networking is a virtual construct. So how are they related? Within a compute host, you have a physical network interface card, and you virtualize that card. We give you this virtual NIC. And that virtual NIC is placed inside the subnet.
And that's the association for the compute host. And that's where the private IP for the compute host comes from, because every compute host or VM you are running, or a bare metal machine, has a private IP address. 
Now, there is another set of dependency the compute instances have, and that's to the boot volume and the boot disk and the block volumes. 
08:42
Lois: What does that mean, exactly? 
Rohit: Well, each of these compute hosts you are spinning up has an operating system. And the image that's used to launch an instance determines its operating system and other software. So you have this concept of an image that comes from this network storage disk called a boot disk. So it doesn't stay on the compute host, it's actually living on the network somewhere. 
And you also have data, like file systems, etc. You're working on the compute instances. They also live on the network. So there is the data disks and operating system disks together. There's a service called block volume service which the compute host uses to run its operating system and run its data disks. Now, these are remote storage. 
09:33
Rohit: There is one more feature which is really relevant when you are talking about compute instances, and that's live migration. We know that computers fail all the time. So how do we make sure that whatever compute host you are running is always up and running, itself? So we have this feature called live migrate. And the idea here is if one of the compute hosts goes down, there's a problem, we would migrate your VM to another host in our data center, and it will be transparent to you. There are multiple options you provide-- whether opt-in or opt-out-- you can choose from. But the idea is we migrate your virtual machines so you can live-migrate between hosts without rebooting. This keeps your applications running even during maintenance events. To achieve this in your own data centers is a not-so-trivial task, but we make that seamless within OCI. 
10:22
Nikita: Thanks for that, Rohit. To learn more about OCI, please visit mylearn.oracle.com, create a profile if you don’t already have one, and get started on our free OCI Foundations training. 
Lois: You will find skill checks that you can take throughout the course to ensure that you are on the right track.
Nikita: We hope you enjoyed that conversation. Join us next week for our final throwback episode. Until then, this is Nikita Abraham...
Lois: And Lois Houston, signing off!
10:54
That’s all for this episode of the Oracle University Podcast. If you enjoyed listening, please click Subscribe to get all the latest episodes. We’d also love it if you would take a moment to rate and review us on your podcast app. See you again on the next episode of the Oracle University Podcast.