Principle of Cloud Computing
(314261)
Week
11
Dr.Hassan Al-Sukhni
PRINCIPLES OF
CLOUD COMPUTING
Module Two Part 3
Src: www.favpng.com
DATACENTER CHALLENGES
3
Challenge 1: Cooling data centers
Cooling plant at a Google DC in
Oregon
4
Challenge 2: Energy Proportional Computing
Average real-world DC and servers are Sub-system power usage in an x86 server as the compute
too inefficient.
load varies from idle to full (reported in 2012).
- waste 2/3+ of their energy
Energy consumption is not proportional
to the load
- CPUs are not so bad but the other
components are.
- CPU is the dominant energy
consumer in servers – using 2/3
of energy when active/idle.
Try to optimize workloads
Virtualization and consolidation. src: “The Datacenter as a Warehouse Computer”
5
Challenge 3: Servers are idle most of the time
For non-virtualized servers 6-15%
utilization
Server virtualization
can boost to
an average 30% utilization
Need for resource pooling and
application
and server consolidation
Need for resource virtualization
src: Luiz Barroso, Urs Hölzle “The Datacenter as a Computer”
6
Challenge 4: Efficient monitoring
Even with virtualization and software
defined DC, resource utilization can be
poor.
Need for efficient monitoring
(measurement) and cluster management.
Goal to meet SLOs and SLIs.
Job’s tail latency matters!
src: “Heterogeneity and dynamicity of clouds at scale: Google trace analysis” SoCC’12
7
Challenge 5: Size and growth of DC (2016-2020)
The scale and complexity of DC
operations grows constantly.
By 2020, Cisco estimated that we would
have 600 million GB of new data saved
each day
(200 million GB big data)
So the volume of BigData by 2020 was
estimated to be as much as all of the
stored data in 2016.
8
Challenge 6: networking at scale
9
Challenge 6: networking at scale (cont.)
Building the right abstractions to work for a range of workload at hyper-scale.
Software Defined Networking (SDN)
Within DC, 32 billion GBs will be transported in 2020
src: Cisco’s report 2016-2020
“Machine to machine” traffic is orders of magnitude larger than what goes out on the Internet
Src: Jupiter Rising: A Decade of Clos Topologies and Centralized control in Google’s Datacenter
Network (ACM SIGCOMM’15)
10
CLOUD COMPUTING DATACENTERS
11
Data Center Technology
Grouping IT resources in close proximity with one another allows for power saving, higher
efficiency in sharing resources, and improve accessibility for IT personnel. The following
issues are concerned:
1. Virtualization
2. Standardization and Modularity (enable economy of scale)
3. Automation (self-configuration, recovery)
4. Remote Operation and Management
5. High Availability (through redundancy)
12
Data Center Technology
6. Security-Aware Design, Operation, and Management (outsourcing resources)
7. Facilities (power, cabling, cooling, fire protection,…)
8. Computing Hardware (standardized commodity servers)
9. Storage Hardware (array, hot-swapping, storage virtualization, fast data replication,
SAN, NAS,…)
10. Network Hardware
Carrier and External Networks Interconnection
Web-Tier Load Balancing and Acceleration
LAN Fabric
Network Gateways
13
Virtualization in datacenters
The common components of a datacenter working together to provide virtualized IT
resources supported by physical IT resources.
14
Summary
A datacenter is a specialized IT infrastructure that houses centralized IT
resources like servers, databases, and software systems.
Data center IT hardware is typically comprised of standardized commodity
servers of increased computing power and storage capacity, while storage
system technologies include disk arrays and storage virtualization.
Technologies used to increase storage capacity include DAS, SAN, and NAS.
Computing hardware technologies include rack-mounted server arrays and
multi-core CPU architectures, while specialized high-capacity networks
hardware and technology, such as content-aware routing, LAN and SAN
fabrics, and NAS gateways, are used to improve network connectivity.
15
References
In addition to the cross-references provided in the slides.
Some material based on:
Lecture notes from “Scalable Systems for the Cloud” by Prof. Giceva at Imperial.
Lecture notes from “Modern Data Center Systems” by Prof. Zhang at UC San Diego.
Book “The Datacenter as a Computer – An Introduction to the Design of Warehouse-scale
Machines” by Luiz Andre Barroso, Jimmy Clidaras, Urs Holzle.
Talk “Inside Azure Datacenter Architecture” with Mark Russinovich (Azure CTO).
Paper “Above the Clouds: A Berkeley View of Cloud Computing”.
Web pages from Amazon AWS, Microsoft Azure, and Google CDP.
16
