7 hours 31 minutes
Hello again. All your cyber and cloud crazed pesky rabbits. This is module six point to where we will learn about the amazing Intel developed technology called hyper threading and virtual and physical CP use. This becomes important as we start spinning up virtual machines in our cloud platform. So having a foundation and virtual ization is very important to your cloud success.
So everyone put on your bunny years and let's hop to it.
Originally, computer processing units had only a single core. That meant the physical CPU had a single central processing unit on it. To increase performance, chip manufacturers needed to improve their processor technology,
and some of those improvements were made by the ability to increase the number. Of course, with the number of central processing units that live on a CPU chip,
a dual core CPU has to central processing units, so it appears to the operating system is to CP use. A quad core processor has four C p used to appear in the operating system eight course eep use and so on. So improvements in the processing manufacturing of the chips
and through miniaturization we now can have in the intel I ate up to his many as 24 28 CPU cores
on a single processor,
so until continues today to crush it in processor development. But back as early as 2000 to Intel, developed hyper threading technology on it. See on X 86 server processors and Pentium four desktop CP use simultaneous multi threading, abbreviated. Hyper threading is Intel's improvement on parallel processing
or doing multiple computational tasks at once.
A processor with hyper threading technology consists of two logical processors for each physical core processor. So what multi threading means is that we have two logical processors sharing the execution engines, the caches and the system bus interface of the physical core by sharing the resource is of the logical processors.
The two logical processes work more efficiently
than the physical core processor by itself. So if one of our logical course stalls on a stall is a waiting period for the rest of the process and the computer to catch up, the other logical course steps in and introduces its instruction. Hyper threading works by duplicating some of the capabilities in the core processor and then creating the logical processor.
This allows the operating system to see what appears to be two processors.
So then the operating systems can schedule two threads of processes simultaneously.
So if you're like me, I'm often double booked and even triple booked in my calendar. And I'm here to tell you that my warped and twisted brain is a single core processor running on a legacy single socket motherboard are really do need hyper threading technology so I can have a logical CPU run each of my double book meetings.
I probably still won't get anything done, but my efficiencies in scale will look really good to management
processors. Calendar is in millions of a second sir microseconds on a micro second toe. A processors like an hourly appointment in our calendar day. Sometimes in the Processors Day, nobody shows up for the appointment. You log into your virtual meeting and no one shows up for your presentation.
This is because everyone is working on other tasks and can't free up to attend our meeting. This is for a processor, a stall. So during the processors meeting of one microsecond where the system stalls and no one shows up,
no new instructions can be given. Well, here comes the logical processor, the second threat of processing instructions that can be performed during our stalled out microsecond now are logical Processor can instruct the system with the new task during our microsecond, while our physical processor is stalled and is timed out, we don't waste our maker Second.
Now we're taking advantage of the idle time to introduce a new processing instruction set.
This is how hyper threading works.
Early processors measured in thousands of a second but now processor's operating microseconds, or one million's of a second, so a little bit dated by today's standards. But until I 76 core processor from a few years ago operated at 31.4, instructions for every one millionth of a second
and these multi core processors with hyper threading turned on by default in the BIOS
gain efficiencies in computing. These instruction sets so hyper threading acts as a bonus for processing but doesn't actually replace physical core processing. While the original consumer processors with hyper threading only had a single core that masqueraded as multiple cores, modern intel CPU is now have both multiple cores and hyper threading technology.
Your dual core CPI with hyper threading appears is four cores to your operating system.
Well, your quad core CPI with high parade hyper threading appears, is a course. Hyper threading is not a substitute for additional course, but a dual core CPI with hyper threading will perform better than a dual core CPI without hyper threading. You will see on the slide that the computer
is seeing a four core processor, but actually it looks like eight logical processors do. The hyper threading technology and what we provisioning our virtual environments in our own data centers in our cloud platforms,
our physical core processors, and are logical core processes. We turn both of these and our software hyper visors when we virtualized our computer environment environment into these things called virtual CP Use or VC pews.
So when provisioning a server and hyper visor, the hyper visor looks at the CPU and determines of hyper threading has supported the picture. On the top left shows how the hyper visor will assign a V, C, P or virtual CPU and one for one manner based on the physical cores or the PCP use.
They're on the motherboard. All servers have to sock it motherboards, meaning there is a physical Intel or a M D processor in each socket.
So when we don't have hyper threading and we have a two sick socket system running quad core processors, we have eight physical cores, or PCP. Use. These physical see pews are then map to eight V CPU so that we can divvy out our virtual machines. Now on the right, we have hyper threading. So in this case we see a two for one relationship
because the hyper visors see on this four core processor
eight logical cores. So RV CPU Zahra signed out for each hyper threat enabled logical corps, not on the physical core. So it's because of hyper threading that we can double our physical core count when we virtual eyes are VC Pease Air. Now, based on the logical course
created by our hyper threading enabled multi core processors, my primary role in network is as an enterprise architect and the design of networks in support of great manufacturers like Cisco
and as I T professionals who understand virtual ization. We need to care about cores and logical cores, v, C, PS and PCP use when we're provisioning standing up and supporting our organization's virtual machines I want you to look at the middle row of boxes where you see Cisco indicating that for their virtual machines, a V C P was directly associated with the P C P U.
Physical cores are required for their virtual machines.
And hyper threading, though, creates great efficiencies and gains and processing power. While Cisco doesn't support hyper threading when provisioning their virtual machines. So when we are in the cloud and we're turning up a new virtual machine in our cloud platform, we have choices that will allow us to provision R V C P is either on logical cores or on physical course.
And we need to care about when we want these applications to run as virtual machines
that they're supported by the manufacturers. On the other hand, there are a lot more applications that are supporting in hybrid threating virtual environments. In this example, when running mission critical applications like voice and video, Cisco wants to be sure
there we can count on processing power to always be there and always dedicated to performing as expected. But most of our applications air not mission critical and so hyper threading is fully supported by the manufacturer
And if we get twice the cores out of our host because we have hyper threading enabled, we can stand up twice the number of virtual machines. So we get a lot more out of our virtual ization investment.
Another topic on this whole V c p u p c p u. Thing is, we have to be careful about over provisioning so we can, over a provision under provisioned and then when running our virtual machines, we could be under an oversubscribed. So what does all this stuff about and what do we mean? I want you to think about your active directory server that you authenticate to every morning when you start your business day,
so you're a D servers really busy at 8 a.m. When all the employees log in at the same time.
But then when you think about it, unless your computer times out or you step away from a long period of time, you don't have to log back in any time soon. This is where your CPU performance and idle times come back in. You want to build your V EMS for the peak times for heavy traffic, but you also don't want over provisions.
So you have all of this computer processor capability just sitting there idle, doing nothing and, frankly, wasting resources and wasting compute dollars. So virtual ization allows youto over subscribe and fill these valleys with
other virtual machines, sharing the same VC pews so you get higher overtime CPU utilization. You want 70 to 80% CPU utilization over time, not just 3 to 7%.
We often focus on all idle instances of our virtual machines waste money. But sizing your VM instances incorrectly is just as detrimental to your monthly bill. It's easy to overspend on large VM instances that air over provisioned for what you really need with any clown service. Whether it's a W S is your G c P. You should always right side your instances in V. EMS
by picking the instant size that is optimized for the size of your workload. Be a compute optimized,
memory optimized or storage optimized. Once your instance has been running for some time, you'll have a better idea whether or not you've chosen the right size and its optimal. Review your usage and may cost estimates with A W s management consoles
Amazons Cloudwatch an AWS trust advisor. If you're using a dip, US as your users can review their metrics for measure, monitor
and Google users can import G C P metrics data for G C P. Virtual machines use this information to find underutilized. Resource is that can be resized to better optimize costs.
So don't run away like a scared rabbit just because it's time for a learning check. No, get back here and grab a carrot. Let's do this.
What is hyper threading?
How many V C PS do we get from a quad core processor when hyper threading is not supported?
What are some of the ways we can provision or virtual machines incorrectly?
And what are three tools that we can use in the clown to right size our virtual machines. So go check the borders of your rabbit pen to see if there's any way out. And if you're still stuck here in this virtual ization lecture, let's review our answers. Do you can get back to all your rabbit habits that are a lot more fun?
So hyper threading is that technology created by until that improves parallel processing so we can do multiple computational tasks At the same time, hyper threading improves parallel processing so much we can actually double or CPU computational capabilities when we virtual eyes
when we are either running absent, don't support hyper threading or running a processor where hyper threading has been disabled. In the BIOS,
we get one virtual CPU for every physical CPU. So with a quad core processor with HT disabled, we would have four VC pews. Weaken. Miss provisions are virtual machine. Several ways we will under provisions are instances of computing memory and storage. So we have poorly running virtual machines.
Rule over provisions are instances and waste valuable computing time
as their CPU sits idle, wasting money and doing nothing. There are a lot of tools in the cloud to make sure we're right sizing RV EM's a W s s trusted adviser and has your has your monitor. And with G C p. We can import and look at our metrics to make sure we're provisioning. RV aims correctly and not underperforming or over performing and thus
overpaying for our cloud service.
So today we learned like a bunch of pesky rabbits what hyper threading is and how we with virtual ization, can double the amount of C P use we can provision from our processors. We spent some time to understand the differences between physical and logical CP. Use VC pews and PCP use and why we care about these differences to make sure abso supported by our manufacturing vendors.
And we learned about the importance of rightsizing our virtual machines.
So next time when we get together, we will close out module six with virtual ization in the cloud. Specifically, it's not just virtual machines for computing. Workloads weaken virtue. Allies weaken virtual eyes. Pretty much anything that moves our packets around from virtual firewalls and virtual routers to virtual switches and wireless land controllers will be talking about standing up in the cloud is a service.
Virtual versions of the same physical devices that are inside our network
really need stuff.
So on behalf of all of the rascally rabbits at the Cyber Security and I T learning team, we want to say thank you so much for joining us. We want to wish you and all of the users on your network farewell. Thank you. Take care and happy packets