CHAPTER 4 CLOUD COMPUTE

CHAPTER 4 CLOUD COMPUTE.

COMPUTE INTRODUCTION. CLOUD COMPUTE. 01. CPU CAPABILITIES.

ENERGY SAVINGS. CLOUD COMPUTE. 06. DEDICATES VS. SHARED COMPUTE.

POLICIES. CLOUD COMPUTE. 11.

01. COMPUTE INTRODUCTION.

CLOUD COMPUTE DEFINED. Computer calculations in the cloud.

CLOUD COMPUTE BENEFITS. Dynamic performance improvement.

CLOUD COMPUTE CHALLENGES. Latency of results - Traffic has to cross the Internet and back.

CONCLUSION OF COMPUTE INTRODUCTION. Cloud compute is computer-based calculations and processing in the cloud.

02. CPU CAPABILITIES.

CENTRAL PROCESSING UNITS (CPUs). The core compute engine.

CPU TECHNOLOGIES. Hyperthreading Multiple threads of concurrent operation Results in multiple virtual CPUs For example, a 4-core hyperthreaded CPU=8 virtual CPUs.

OVERCOMMITMENT RATIOS. Utilize real resources for virtual machines well Scenario: • 2 CPUs • Each is quad core • Each is hyperthreaded • Total of 16 virtual CPUs • Run 4 virtual machines, each with 8 CPUs • Result is a 2:1 overcommitment ratio - 2 virtual processors for each of the CPUs (including hyperthreading) Overcommitment is the primary factor in private clouds Cloud service providers hide this from you and perform it themselve.

CONCLUSION OF CPU CAPABILITIES. The CPU is the core compute engine and cloud providers offer varying levels of capability.

03. MEMORY REQUIREMENTS.

W HAT ARE THE FACTORS OF MEMORY REQUIREMENT?. 1. Operating system.

MEMORY BALLOONING. A feature of virtualization platforms • Unused, allocated memory for one guest can be used by another • Allows for overcommitment of memory Mostly used in private clouds from a configuration perspective • Service providers may used it, but you won’t configure it Bursting • The action of ballooning.

CONCLUSION OF MEMORY REQUIREMENTS. Consider everything running on a virtual machine when determining memory requirements.

04. PERFORMANCE CONSIDERATION.

PERFORMANCE FACTORS. CPU. MEMORY. DISKS. NETWORK.

DEMO. Optimizing CPU options in AWS https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instance-optimizecpu.html.

CONCLUSION OF PERFORMANCE CONSIDERATION. CPU and memory are the primary performance factors specifically for compute.

05. COST CONSIDERATIONS (LAB).

HANDS-ON. Azure Pricing Calculator https://azure.microsoft.com/en-us/pricing/calculator.

CONCLUSION OF COST CONSIDERATIONS (LAB). Costing in the cloud is a combination of desired resources and performance requirements.

06. ENERGY SAVINGS.

PUBLIC / COMMUNITY CLOUD ENERGY SAVINGS. Shared resources = Energy savings.

TRADISIONAL PRIVATE DEPLOYMENTS. Departmental servers.

PRIVATE CLOUD DEPLOYMENT ENERGY SAVINGS. Virtualization changed everything Private cloud is basically automated virtualization • With some extra bells and whistles The new deployment that saves energy • Multiple virtual servers on a single box • Services accessed across the Internet in the private cloud • Virtual desktops with high computing power - Possible shared among multiple resources.

CONCLUSION OF ENERGY SAVINGS. Shared resources in the public and community clouds results in overall energy savings.

07. DEDICATED vs. SHARED COMPUTE.

DEMO. AWS Dedicated Hosts and Instances https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/dedicatedhosts-overview.html.

CONCLUSION OF DEDICATED vs. SHARED COMPUTE. A dedicated host is a physical server that is used only for your instance(s).

08. HIGH AVAILABILITY AND DISASTER RECOVERY FOR COMPUTE.

HA/DR EFFECT FOR COMPUTE. High availability and disaster recover (HA/DR).

HA/DR EFFECT FOR COMPUTE. Availability functions for compute Clustering - Multiple instances with a primary and failover Load balancing - Multiple instances with workload rotating between them Serverless limits - Imposed by service provider.

DEMO. AWS Lambda Limits https://docs.aws.amazon.com/lambda/latest/dg/limits.html.

CONCLUSION OF HIGH AVAILABILITY AND DISASTER RECOVERY FOR COMPUTE.

09. MONITORING (LAB).

MONITORING TERMINOLOGY. TARGET OBJECT. ALERTS. Baselines Anomalies.

EVENT CORRELATION. Event timestamps are used to correlate Ex: • Event A happened at 10:17:32 and Event B happened at 10:17:33 • Event B and A are related • Maybe Event B was caused by Event A Correlation benefits • Determination of cause • Locating attack points • Identifying errant code.

CONCLUSION OF MONITORING (LAB). An anomaly is an event outside of the ordinary expectations.

10. FORECASTING.

KEYPOINTS OF FORECASTING REQUIRED RESOURCES. Forecasting is looking into the future to determine needs - Look at today to predict tomorrow.

Upsize/increase or downsize/decrease resources to meet future demands - CPU - Memory - Storage.

CONCLUSION OF FORECASTING. Forecasting is about looking at yesterday and today to predict tomorrow.

11. POLICIES.

KEYPOINTS OF POLICIES AND MONITORING. Monitoring may reveal sensitive data - Can monitor down to the process level - Identifying processes can give you insight into points of attack.

Policies in support of event collection - What can be monitored? - When should it be monitored? - What can be correlated?.

CONCLUSION OF POLICIES. Monitoring can reveal sensitive data.