Key Points discussed in this webinar are:
1.How dynamic simulation can replace traditional network simulations that are slow and lack configuration and visibility to analyze performance.
2.How to avoid over or under design, cost increases, and delays.
3.How an architectural model can be used to test the capacity and power requirements of your data center or your server.
Contact us at info@mirabilisdesign.com for any queries.
Capacity Planning and Power Management of Data Centers.
1. SAVING COSTS, REDUCING POWER AND
ELIMINATING EXCESS CAPACITY
IN
DATA CENTERS
Deepak Shankar
Founder
Mirabilis Design Inc.
Email: dshankar@mirabilisdesign.com
2. Logistics of the Webinar
2
To ask a question, click on Cloud Chat sign and type the
question. Folks are standing by to answer your questions.
There will also be a time at the end for Q&A
3. SAVING COSTS, REDUCING POWER AND
ELIMINATING EXCESS CAPACITY
IN
DATA CENTERS
Deepak Shankar
Founder
Mirabilis Design Inc.
Email: dshankar@mirabilisdesign.com
4. Agenda
Introduction to Data Center design requirements
Case Study 1: Sizing the data center and estimating performance and power
About Mirabilis Design
5. NEW APPROACH TO DESIGNING DATA
CENTERS
Deepak Shankar
Founder
Mirabilis Design Inc.
Email: dshankar@mirabilisdesign.com
6. Sizing Data Centers Today
Current and
Future SLA
Get Current Configuration
with usage estimates
Add more servers
for future needs
Create configure using
experience, articles
and references
Try to fit more
customers
7. Challenges with Current Approach
Budget planning is difficult
◦ Limited visibility into operating expenses and system utilization prior to usage
Capital expense is based on “build-and-they-will-come” model
◦ Need to invest in expensive equipment, software licenses and performance measurement software
Scalability means adding more servers
◦ Sharing requires good quality load distribution based on anticipated work loads
Missing out on revenue opportunities
◦ Reuse floor space, power connection and existing hardware
8. Evaluating Data Center
Hardware and Software
Scheduling/Arbitration
proportional
share
WFQ
static
dynamic
fixed priority
EDF
TDMA
FCFS
Communication Templates
Architecture # 1 Architecture # 2
Computation Templates
DSP
AI
GPU
DRAM
CPU
FPGA
m
E
DSP
TDMA
Priority
EDF
WFQ
RISC
DSP
LookUp
Cipher
AI DS
P
CPU
GP
U
mE DD
R
static
Which architecture is better suited
for our data center?
9. Migrate to a Reliable and Accurate
Method of Data Center Design
Simulate the theoretical and analytical studies for network topologies, traffic distributions,
scalability, and load balancing
Graphical prototype of the proposed data center or colocation site
Setup attributes for the communication, computation, storage and software
Analyze the measured statistics and reports
System sizing to meet SLA and profitability requirements
Optimize to ensure highest efficiency
10. Concept of a Server System Modeling
Applications are a set of Complex tasks
- Variable rate input stream and task
dependency
Contention between resources and tasks
- Task assignment and sequence
Trade-off between process and transfer
- Reduce buffering and latency with minimal
transfer time
I/O
DSP
CPU1
CPU2
task1 task2 task3 task4
Scheduling software tasks using limited resources
11. End-to-end Application Evaluation
Device 1
Analog
Front-
End
RISC-V
core
Modem
(BLE,GPS
WiFi, 5G)
Radio
Transceiver
Network
Hub Data
Center
Model the Source, Internet/Wireless and Data Center for true system evaluation
12. Use Generated Reports to Make Decisions
Reports can cover network latency, hardware efficiency, power, reliability and throughput
13. Advantages with Data Center Modeling
Gain valuable insights into the data center operations before deployment
◦ System efficiency, operational cost, Bill-of-materials
Projections for investments based on SLA, workloads and migration to cloud
◦ Get accurate measurements for each configurations
Simplify system complexity
◦ Explore consolidation using VM, reduce device count and try different load distribution algorithms
WorldView
◦ Simulating multiple data centers and distribution traffic by distance, dynamic usage and response time
14. Key terms in System Modeling
Architecture Exploration
◦ Optimize and validate the system specification
◦ Specification: Processor speed, topology and arbitration
◦ Requirements: Timing, energy, cost, weight and efficiency
Performance Analysis
◦ Buffer size, utilization, throughput and response time
Power Measurement
◦ Peak and average power, energy and power/task
Functional Correctness
◦ Arbitration, software task scheduling and task graph
Failure Analysis
◦ Hardware, Software, network, data, power and logic
Making Better Quality Products
15. Sizing Data Centers of the Future
Current and
Future SLA Data center model of
Servers, storage,
software and interfaces
Simulate Requirements
of the future
Optimize for energy
cost, latency and
throughput
Feedback changes in
traffic, applications and
security
16. USE CASE 1:
SCALABLE DATA CENTER
Deepak Shankar
Founder
Mirabilis Design Inc.
Email: dshankar@mirabilisdesign.com
17. How does Modeling and Simulation of
Data Center Architecture Work?
Use SLA and Future plans
◦ Goal input for the model
Construct a model
◦ Communication: Interfaces and multiple data center locations
◦ Hardware: Storage, processor (cores), memory
◦ Inter-process communication: Task graph to show the relation between requests, applications and mapping to hardware
◦ Software: VM,OS, Application
◦ Licenses: Number of instances
Add workload and application
◦ Add the traffic stimulus as traces, distribution or custom patterns
View reports
◦ Match metrics against the goals to select the most suitable configuration
Test
◦ Failures, traffic overload and loss of hardware/interfaces
18. Datacenter modelling – Block diagram
Interface 1
Interface 2
Interface 3
Interface n
.
.
.
OS1
OS2
OS3
OSn
Task
lookup
Hypervisor
Proc
Core1
Proc
Core2
Proc
Core3
Proc
Core4
Proc
Core n
SSD1
SSD2
SSD
n
HDD
1
HDD
2
HDD
n
Request
1
Request
2
Request
n
Datacenter
19. Datacenter system - Overview
Traffic profile : Wireshark
Number of Parallel flows : 3
Number of Interfaces : 5
Interface speed GHz :3200
Number of OS : 5
Number of Processors : 15
Number of cores per processor : 16
Number of SSD : 500
Size of each SSD : 1 TB
Number of HDD : 500
Size of each HDD : 1 TB
24. About Mirabilis Design
Founded in 2008 and based in Sunnyvale, CA, USA.
Development and support centers in US, India, China, Japan and Czech Republic
System design of data centers, electronics, semiconductors and software
VisualSim- Modeling and simulation software
First platform for rapid prototyping of data centers
100’s of man years experience in system design and exploration of complex design
Select the “Right” configuration to match customer request
25. Application of VisualSim
Monte-Carlo simulation with random samples,
parameters, connectivity, traffic and use-cases
Models constructed with library of pre-defined
parameterized components
◦ Resource, custom development and HW/SW/NW
Graphical and hierarchical construction,
debugging and analysis of model
Batch-mode simulation for large-scale analysis
and experimentation
Interfaces to languages, simulators and
spreadsheets
25
Performance
Analysis
Power
Exploration
HW-SW
Partitioning
Failure
Analysis
Software
Network
Hardware
Validate and optimize your design quickly and accurately
26. VisualSim software with libraries
Training:
Training and modeling support- user builds
the components and models
Services:
Develop custom library- User assembles
the models
Develop custom libraries and models -
User conducts parameter study
Architecture evaluation- Will develop
model, analyse and provide feedback
Model-based Systems Engineering simplified and made easy-to-adopt
Mirabilis Design Software and Solutions
27. SAVING COSTS, REDUCING POWER AND
ELIMINATING EXCESS CAPACITY
IN
DATA CENTERS
Deepak Shankar
Founder
Mirabilis Design Inc.
Email: dshankar@mirabilisdesign.com