BITS Pilani

presentation
BITS Pilani WILP
Pilani Campus
 BITS Pilani
Pilani Campus

CSIWZG522 Design and operations of Data center

CS 04
Books

Text Book(s)
No Author(s), Title, Edition, Publishing House
T1 Building a Modern Data Center: Principles and Strategies of Design
by Scott D. Lowe (Author), David M. Davis (Author), James Green (Author),
Seth Knox (Editor), Stuart Miniman (Foreword)

Reference Book(s) & other resources

No Author(s), Title, Edition, Publishing House
R1 Data Center for Beginners: A beginner's guide towards understanding Data
Center Design
B.A. Ayomaya (Author)

BITS Pilani, Pilani Campus

Advisory

• The printed book may or may not be available in the

market.
• The ebook is available in the internet.
• Students are strongly advised to make notes from RLs &
CSs for examination purpose right from the beginning.
• Unavailability of the text book will not be an excuse to
allow print copies of the ebook or PPTs during
examinations.

BITS Pilani, Pilani Campus

Slide references

Building a Modern Data Center:

Principles and Strategies of Design
by
Scott D. Lowe (Author),
David M. Davis (Author),
James Green (Author),
Seth Knox (Editor),
Stuart Miniman (Foreword)

BITS Pilani, Pilani Campus

Lecture Plan
Contact Hour 7&8

Time Type Description Content

Reference

Pre CH RL 2.2 Computing components of Data center

During CH CH2 The No-Spin Zone: The Move from Disk to T1: Ch-2
Flash
The Fall of the Monolithic Storage Array
The Emergence of Convergence
The Role of Cloud
Cloud Types
Cloud Drivers

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash
• Magnetic storage media has been the dominant choice for data storage for the
majority of data center history.
• Spinning disks have served as primary storage, and tape-based storage
systems have served higher capacity longer term storage needs.
• However, the performance of spinning disk eventually leveled off due to physics-
induced limitations.
• The speed by which data on a spinning disk can be accessed is based upon a
few factors, but the one that is the biggest problem is the rotation speed of the
disk platter.

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash
• Eventually, the platter can’t be spun any faster without damaging it.
• Based on what the storage industry has produced in the past few years, it would
appear that 15,000 rotations per minute (15K RPM) is the fastest speed that
manufacturers have been able to maintain while keeping the disk economically
viable to the customer.
• A 15K SAS drive is a high-performing disk, to be sure.
• However, the number of I/O operations that any spinning disk can perform in a
second doesn’t seem to be changing all that much.
• The fastest, most efficient spinning disks can deliver less than 200 random I/O
per second (IOPS).

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash
• While this is beyond adequate for a use case like a PC, it has left something to
be desired when serving I/O to a dense, mixed workload virtual server or virtual
desktop environment.
• The numbers get even trickier when RAID write penalties are factored in;
depending on the RAID configuration a number of disks may be needed to
achieve 200 IOPS rather than just one.
• There’s also the issue of latency.
• Due to the mechanical nature of a spinning disk drive, latency (the time it takes to
retrieve or write the data in question) can’t be pushed below a certain threshold.
• Tiny bits of latency added together across many drives becomes an issue at
scale.

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash
• The solution to both the IOPS problem and the latency problem is found in flash
storage.
• In short, flash storage media makes use of non-volatile memory to store data as
opposed to magnetic platters.
• Although the use of flash storage was initially troublesome due to durability
issues, the performance has always been quite attractive and often worth the
risk.
• Because flash storage is not mechanical in nature, it doesn’t suffer from the same
limitations as spinning disks.
• Flash storage is capable of latency on the order of microseconds as opposed to
spinning disk’s multiple milliseconds.
• It’s also capable of far more I/O operations per second than a handful of spinning
disks.

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash

• The issue of durability has been solved over time as manufacturers

improve the physical memory, storage controllers use intelligence like wear
leveling, and different types of flash cells are developed, like single level
cell (SLC), multi-level cell and enterprise-grade multi-level cell
(MLC/eMLC), and triple level cell (TLC).
• A typical eMLC drive on the market in 2015 is warrantied for 10 full writes
per day over a period of 5 years.
• Alternatively, some manufacturers specify simply a total amount of data
written.
• The same eMLC drive would probably be warrantied for something like 3.5
PB of data written.
• Lastly, because of the non-mechanical (or “solid state”) nature of flash
storage, it requires much less power to operate when compared to spinning
disk.

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash
• As data center power bills have always run high, any way to reduce power
consumption is attractive to the data center manager — and the CFO! In some
countries, governments offer substantial incentives for making environmentally
friendly changes like reducing power consumption.
• In some cases, purchasing boatloads of flash storage to reduce power
consumption may be cheaper than the cost of the fine for failure to comply.
• Flash storage becoming widely available has been a huge win for the data
center industry.
• It allows much higher performance with substantially less power, and in the
future flash storage will be available at a cost per gigabyte comparable to that
of spinning disk.

BITS Pilani, Pilani Campus

The No-Spin Zone: The Move from Disk
to Flash

• Exactly how long this will take is anyone’s guess, but industry experts predict
it could take until at least 2020.
• This maturing of flash storage has led data center architects to reconsider
the way storage is accessed in the data center yet again.
• Just as the utilization and management issues of direct-attached storage
gave birth to the monolithic storage array, performance issues and
power/environmental concerns have birthed a new storage design.
• The data center of the future will likely see less of the monolithic storage
array in favor of a return to direct attached storage ... but with a twist.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• Monolithic storage arrays solved many of the data center’s problems and
allowed IT to achieve greater efficiencies and scale.
• Unfortunately, the things that made this architecture so attractive also
eventually became its downfall.
• The virtualization of compute led to densities and performance requirements
that storage arrays have struggled to keep up with ever since.
• One of the primary challenges that manufacturers of monolithic storage
arrays have been trying to solve for a number of years is the challenge of the
“mixed workload.”
• By the nature of virtualization, many different applications and operating
systems share the same physical disk infrastructure on the back end.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• The challenge with this architecture is that operating systems, and especially
applications, have widely varying workload requirements and characteristics.
• For example, attempting to deploy virtual desktop infrastructure (VDI) on the
same storage platform as the server virtualization has been the downfall of
many VDI projects.
• Due to the drastically different I/O characteristics of a desktop operating
system versus a server operating system and the applications running on
them, they require almost completely opposite things.
• An average Windows server might require 80% reads and 20% writes,
whereas on the exact same storage array, with the same disk layout, same
cache, and so on, a virtual desktop might require 20% reads and 80% writes.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• Couple this problem with the fact that hundreds — perhaps thousands — of
virtual machines are trying to perform these operations all at the same time and
you have what the industry has dubbed “the I/O Blender.”
• This is a comical metaphor, but quite accurate at describing the randomness of
I/O operations coming into the array.
• As application performance requirements go up, it has also become
increasingly important to provide very low latency.
• So which storage model is likely to have lower latency: the one where storage is
accessed across a network and shared with all other workloads, or the one
where storage is actually inside the server doing the processing on the SATA/
SAS or PCIe bus, or even in memory?
• Of course, the answer is the model where the storage is local to the workload.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• Bus speeds versus network speeds are on totally different orders of
magnitude.
• With that in mind, some new ideas have started popping up in the data center
storage market over the past few years.
• Figure 2-5 shows the progression of storage design over time.
• One idea is the concept of server-side caching.
• This design is less radical than others, because it continues to make use of
existing shared storage.
• One of the first really well done implementations of this technology in the
enterprise was a solution that used a virtual machine to consume local DRAM
and use it as a high-speed cache.
• Another early option was a very expensive but high-performing PCIe SSD
that accelerated remote storage with local caching.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array

• These designs solved common problems like boot storms in VDI

environments, because the virtual machines on each host were able to
retrieve hot blocks from local DRAM before ever traversing the network.
• This technology was mimicked and improved on, and today a number of
options exist for caching with local SSDs and DRAM in front of shared
storage arrays.
• A different, more radical architecture is becoming more common, however,
and will allow IT organizations to replace and/or pool the monolithic
storage arrays for general purpose workloads in the future.
• This design, which will be discussed in-depth in a later chapter, is enabled
by software defined storage (SDS).

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• The data center of the future looks (physically) a lot more like the data center of
the past, in which a number of servers all contain their own direct attached
storage.
• The difference is that all of this locally attached storage is pooled, controlled,
accelerated, and protected by a storage management platform running on the
hypervisor.
• Local storage is just a bunch of SSD disks rather than being configured in a
RAID group, and fault tolerance is controlled at the node level rather than at
the storage controller level.
• The resilience could be thought of like a network RAID, although it’s more
complex than that.

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• The performance and scale implications of this model are massive: because
each node added to the cluster with local storage contributes to the pool, this
means that the storage pool can grow to virtually limitless heights.
• Each server that is added has its own storage controller, meaning that
throughput never becomes an issue.
• Increasing capacity of the pool is as easy as adding disks to existing servers or
adding more servers overall.
• The control of all of this is done by either virtual machines (VSAs) or by kernel-
level software, and the administrator typically manages it from the hypervisor’s
existing management interface (like vCenter or SCVMM – System Center Virtual
Machine Manager).

BITS Pilani, Pilani Campus

The Fall of the Monolithic Storage
Array
• SDS is changing the data center in tangible ways, and as more organizations
begin to adopt this architecture, vendors of monolithic storage arrays will have
to innovate or pivot in order to stay relevant and survive.
• A deep dive on SDS is coming in a later chapter, but the stage isn’t fully set
yet.
• There’s a lot of moving pieces in the data center and it can get a bit
overwhelming at times.
• Plus, it moves so fast.
• Wouldn’t it be nice if the design and building of it was left to someone else, and
what showed up on the loading dock was ready to use rather than ready to
begin a 6-month project?

BITS Pilani, Pilani Campus

The Emergence of Convergence

• As the challenges for IT have grown in equal proportions with the ever-
increasing scope of their responsibilities, IT decision makers have often looked
to outsource parts of their operation.
• A notable trend for data center “outsourcing” of sorts is now referred to as
convergence.
• Put simply, convergence is multiple pieces of the infrastructure assembled prior
to delivery to the customer.
• Convergence saves time and frustration during the deployment phase and
provides decreased time-to-value after procurement.
• An example of a common form of convergence might look like this: a rack is
delivered to the data center already containing a storage array, a blade chassis
populated with blades, and a few top-of-rack switches.

BITS Pilani, Pilani Campus

The Emergence of Convergence

• Everything is cabled up, and all the configuration of the switching and storage
has been done prior to delivery.
• At the moment the converged stack is delivered, the data center team can roll
into place, deliver power and upstream network connectivity, and the pod will be
up and running.
• This model of growing the infrastructure is substantially faster than the
traditional model of having parts delivered, assembling them, hiring consultants,
troubleshooting, and so on.
• Speaking of troubleshooting, there’s another important facet to this approach:
the pod that comes pre-built is based on a tested and validated reference
architecture.
• This means that the customer doesn’t need to tinker with exactly which
configuration of the parts available will work for them; that design work has
already been done.

BITS Pilani, Pilani Campus

 The Emergence of Convergence

• Also, when the pod is built at the factory, the technician building it actually makes
sure that the connections are good and the infrastructure is highly available and
performing as designed.
• The value in convergence comes not only from the fact that the solution comes
pre-assembled, but also from the fact that it includes all the pieces necessary.
• Half the challenge in traditional piecemeal solution-building is getting all the right
parts and ensuring interoperability.
• Convergence guarantees that with the purchase of a certain SKU [Stock Keeping
Unit], all the components contained within it will be compatible with one another,
and all the necessary parts will be included.

BITS Pilani, Pilani Campus

The Emergence of Convergence

BITS Pilani, Pilani Campus

The Emergence of Convergence

• Convergence has helped many organizations realize project objectives

faster, and has saved a multitude of headaches over time.
• But if a little convergence was good, does that mean a lot of convergence is
great? A design methodology that will be discussed at length in following
chapters is now taking the place of convergence in the data center.
• The successor to convergence is known as “hyperconvergence,” and it takes
the idea of simplicity to the customer to new heights.
• Hyperconvergence is so called because of the scope of what is being
converged (Figure 2-6).

BITS Pilani, Pilani Campus

The Emergence of Convergence

• In a converged infrastructure, many infrastructure components are brought

together into one rack (or a few racks).
• In a hyperconverged infrastructure (HCI), those same components are brought
together within a single server node.
• Hyperconvergence is born from cloud data centers that pioneered and
leveraged this technology to operate at the massive scale they require.
• Now, what’s a technology designed for cloud data centers like Facebook or
Amazon doing in a small corporate data center? It turns out that the cloud isn’t
just for the top 1% anymore.
• Cloud technology is being leveraged all over the world by even small
companies.

BITS Pilani, Pilani Campus

The Cloud
• Keep in mind that the cloud is simply a method of offering and provisioning
on-demand services.
• With this definition in mind, it’s easy to see that a private cloud deployment
is simply an on-premises deployment of a tool like OpenStack that allows for
rapid, on-demand provisioning of resources that can easily be created and
destroyed.
• But why does anyone do this anyway?
• What is the value in cloud-based solutions as opposed to the way virtual
infrastructure has been deployed for the previous decade?
• The Cloud Drivers section answers these questions.

BITS Pilani, Pilani Campus

The Role of Cloud
• The term cloud has always been a bit confusing and hard to nail down
• Unfortunately, many misconceptions exist about exactly what “the cloud” is,
but in the most general sense, the cloud is pretty easy to grasp
• Cloud computing is a model of delivering infrastructure or application
resources in a way that is flexible, rapid, and on-demand
• This is why purchasing infrastructure from Amazon Web Services (AWS), for
example, would be classified as cloud
• It’s on-demand, takes about two minutes to provision, and has tons of
options
• Because cloud is a model and not a thing, there are a number of different
ways cloud infrastructure can be implemented.

BITS Pilani, Pilani Campus

Cloud Types

• Different cloud deployment models fit different organizations.

• There are certain cases where an application has been developed from the
ground up to run in a cloud.
• In this case, it may make sense to use a public cloud model, where all
resources are provisioned in a third party data center provided by the likes of
AWS, Microsoft, VMware, Google, or your friendly neighborhood cloud
provider.
• Especially for some small businesses, being entirely public-cloud-based
allows for an extremely light IT footprint in the office or storefront, resulting in
less overhead.
• Public cloud can be very affordable.
• It also offloads risk and overhead in terms of compliance, patching,
equipment failure, hardware refreshes, and so on.

BITS Pilani, Pilani Campus

Cloud Types

• When purchasing cloud resources, you’re purchasing a service and do not care
what happens on the back end.
• If it works for the organization, an exclusively public cloud model is great — but it
doesn’t work for everyone.
• The next possible choice is a combination of on-premises cloud and public
cloud; it’s known as hybrid cloud.
• Using this model, IT resources run in the corporate data center as usual, but an
extension to a public cloud data center is in place.
• This means that based on certain requirements, constraints, or other design
decisions, a workload can be provisioned either to the private data center or to
the public one.

BITS Pilani, Pilani Campus

Cloud Types

• An example of how hybrid cloud might work is that of a retailer.

• If Black Friday is coming up, the retailer may be able to spin up an extra 20
instances of their website and shopping cart application in the public cloud.
• The back end databases still exist in the on-premises data center and need not
be migrated.
• This is commonly referred to as “bursting” to the cloud.
• Another example where a hybrid cloud model could work out well is in an
organization that has a heavy in-house development workload.

BITS Pilani, Pilani Campus

Cloud Types

• If developers are constantly creating and destroying test environments, it can

require lots of horsepower to keep things running fast enough that developers
are happy, and project scopes can change with a moment’s notice.
• A much easier way to handle this situation would be to run production workloads
in the on-premises data center, but have development and testing workloads
provision to the public cloud.
• This can also save on cost as opposed to running the resources locally.
• The third option for cloud deployment models is a private cloud.
• This phrase can be rather confusing if one thinks of “cloud” as a third party
selling services on the Internet, or worse yet, if one thinks the Internet itself is a
cloud.

BITS Pilani, Pilani Campus

Cloud Drivers

• Although virtualization revolutionized the data center, expectations for

performance, availability, and cost never cease to change the way things must
be done.
• Eventually, the increased speed and reduced cost of virtualization wasn’t
enough anymore.
• There are a few things driving the adoption of cloud models, currently; as with
most decisions, it ultimately comes back to the business and the bottom line.
• Many organizations develop software.
• Some develop it to sell, and software is their product; others develop software
for their own internal use.

BITS Pilani, Pilani Campus

Cloud Drivers

• Either way, developers are the resources driving change in the data center.
• Because the software development lifecycle has become so important to so
many business, any technology or deployment model that will allow that cycle to
iterate faster will be of benefit to the business.
• Therefore, the first driver for cloud models (public, hybrid, or private alike) is
agility.
• By nature, any sort of cloud model will dramatically increase the speed of the
software development lifecycle.
• The second driver is cost.
• Many IT organizations are required to accomplish more projects than last year
with less budget than last year, and they have to look at all available options.

BITS Pilani, Pilani Campus

Cloud Drivers

• In the case of public and hybrid cloud, the cost of running workloads (especially
ephemeral ones) in that way can cost significantly less than purchasing and
configuring the hardware to accomplish the same goal on-premises.
• In the case of on-premises, private cloud, cost savings can be found in the fact
that fewer developers iterating faster will accomplish the same work as more
developers iterating slowly.
• Providing the tools needed to iterate quickly could allow the paring back of
developer resources.
• A third driver is scale. By leveraging a public cloud provider, the scale to which
an organization’s systems can grow is practically limitless.

BITS Pilani, Pilani Campus

Cloud Drivers

• Where physical space and limitations on power, cooling, and other factors
make scale a challenge when hosting the entire infrastructure on-premises,
the public cloud makes scaling a breeze.
• Finally, the cloud infrastructure model is now important because the
subscription-based fee is looked on favorably in budget meetings as
opposed to large capital expenditures.
• The shift to operational expense as opposed to capital expenditure allows for
much more flexibility year to year and even month to month.
• Because of that flexibility and control, many organizations choose cloud-
based models to help move their IT projects forward and stay within budget.
• It’s safe to say that whether public, private, or a mix of both, the use of cloud
infrastructure has changed the way the modern data center does business.

BITS Pilani, Pilani Campus

THANK YOU !!

BITS Pilani, Pilani Campus