VMTurbo enables greater performance for large virtual

VMTurbo enables greater performance for large virtual
VMTURBO ENABLES GREATER PERFORMANCE FOR LARGE VIRTUAL
ENVIRONMENTS
As the number of virtual machines (VMs) in your infrastructure grows, so do the
challenges. Delivering top performance for database VMs can be a tricky, timeconsuming task for administrators. They have to sort through massive amounts of data
and use trial and error to tweak resource allocation daily to get the best results.
VMTurbo Operations Manager is a platform that makes managing VMware
vSphere VMs easier. VMTurbo understands the resource demands for all your
applications, VMs, and containers, and makes decisions to assure they get the resources
they need to perform. These decisions also ensure you use your resources wisely. We
put the promise of VMTurbo to the test in the Principled Technologies labs, and we
found that it did improve performance in our large-scale, 100 VM environment. When
we used VMTurbo decisions, our cluster was able to handle up to 33 percent more
database orders per minute with 23 percent better response times after one week of
discovery.
NOVEMBER 2015
A PRINCIPLED TECHNOLOGIES REPORT
Commissioned by VMTurbo Inc.
THE IMPORTANCE OF OPTIMIZATION
If you buy a new dresser, do you fill only half of the drawers and then purchase
additional furniture so that you can store all of your clothes? It may sound silly, but this
is exactly what happens in datacenters across the globe with servers and storage. After
making careful hardware selections based on ability and capacity, expensive resources
can still be underutilized. An inability to assure application performance drives
companies to purchase more hardware and expand the datacenter. This approach is
costly and wasteful.
Even if your organization has done a good job of right-sizing your infrastructure
for the VMs you need, it’s still likely that processor, memory, and storage resources
aren’t being used efficiently and that performance is suffering from resource contention. Traditionally, administrators have had to look at logs of data to figure out
where inefficiencies occur and then make those tweaks manually to try to optimize the
environment for top performance. However, that process is time-consuming and isn’t guaranteed to find every way to improve your infrastructure.
VMTURBO OPERATIONS MANAGER CAN HELP
VMTurbo Operations Manager assesses system-level resource demands in
VMware vSphere environments automatically and makes automatable decisions that
optimize performance. It discovers what application demands are and where storage,
processor, and memory resources need to move to meet those demands with ease. In
our testing, a pattern emerged. Utilizing the insight the Application Control Module and
Storage Module granted, VMTurbo recognized the I/O demands of the SQL Servers in
the test environment. Figure 1 shows an example of how VMTurbo intelligently placed
virtual disks based on the performance of the underlying storage for some VMs.
Throughout the week, VMTurbo would incrementally move more I/O heavy VMDKs to
flash or high-speed disk-based storage, while shifting low-demand VMDKs to high
capacity, low performance spindles. A key feature to the overall performance
improvement over the solution without VMTurbo was that the VMTurbo decision
engine continuously managed the tradeoff between performance and capacity
improvements against the current workload. This means that storage and VM
migrations happened one at a time, minimizing additional load on the storage and
network subsystems.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 2
Figure 1: How VMTurbo helps optimize performance.
As Figure 2 shows, using the VMTurbo decision algorithms to place the SQL
workload across existing storage, hot-add vCPUs, and determine and provision
additional hardware increased the total OPM output of our 100-VM infrastructure by 33
percent. We performed Week 1 testing without VMTurbo. During this time, Storage DRS
successfully rebalanced storage allocation for more efficient utilization of datastore
resources, while incrementally improving performance through I/O monitoring and
analysis.
During Week 2 (in green), we enabled VMTurbo automation and manually
enacted VMTurbo-prescribed measures which included adding lower-performance RAID
10 SAS storage in the cluster and two additional hosts. VMTurbo optimized the
workload across these new resources automatically, placing the VMs equally on the
identically configured blades. VMTurbo also recognized the storage as lower
performing, and selectively migrated only enough VMDKs to lighten the performance
burden on the original storage pools without degrading the performance of the
relocated VMDKs.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 3
Figure 2: Using VMTurbo to
control and implement virtual
and physical resource
increases to our large-scale
environment resulted in OPM
increases of up to 33 percent
over the week.
In our tests, we ran six different workloads on the VMs: 20GB light, 20GB
medium, 40GB medium, 40GB med-high, and 40GB high. Figure 3 shows the
performance by workload over the course of the week.
Figure 3: Using VMTurbo and
implementing its suggested
configuration changes and
provisioning instructions
increased performance across
all workloads over the week.
A faster system with less latency, or wait time, goes a long way to improving
user satisfaction with applications that run on that server. As Figures 4 and 5 show,
application latency decreased by up to 23 percent on average after implementing a
week’s worth of decisions from VMTurbo, which included automated actions such as
hot-adding vCPUs and memory to resource-constrained VMs, increasing SQL transaction
log sizes, granular VMDK migrations based on specific workload demands, and
provisioning additional disks and hosts when requested by VMTurbo.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 4
Figure 4: VMTurbo decreased
application latency, which
means that users get a faster
response when they complete
their tasks.
Figure 5: VMTurbo decreased
application latency, which
means that users get a faster
response when they complete
their tasks.
WHAT KINDS OF CONTROL DID VMTURBO ENABLE?
With the integration of control modules into VMTurbo 5, administrators now
have a variety of options to extend control across their environment. Each module
provides an expanded understanding of the supply chain of resources that applications
need to perform, enabling decisions that are more comprehensive and the ability to
prevent future problems. In our testing of this environment, we found that storage was
the primary bottleneck on the performance, so most decisions targeted storage issues.
At the most basic level, VMTurbo decides on moving VMs to less-congested
storage. Beyond that, it provides alternatives, such as moving a datastore to a different
pool of disks and provisioning more disks to a pool to alleviate capacity and
performance choke points. A key feature of the VMTurbo automated host and storage
migration is the decision of which VMs to migrate. Rather than migrate a busy VM from
a congested host or storage volume, which will increase latency for the active VM,
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 5
VMTurbo chooses less-utilized VMs, which are less likely to suffer from the migration
process.
Figures 6 and 7 show a timeline of VMTurbo automated actions including
additional storage and host provisioning decisions. The first day, we allowed VMTurbo
to provide decisions based on the previous week’s results, without providing any additional hardware. VMTurbo spent the day migrating VMs to more appropriate
storage (see Figure 1), and provided additional vCPU resources to VMs using VMware
hot-add functionality. The SQL application control module also provided decisions to
increase the vMemory allocation to several VMs, which we performed manually via hotadd.
On the second day of testing, the VMTurbo decision engine requested
additional storage resources due to disk latency and capacity. In response to the
request, we provisioned a new RAID 10 pool of eight 15K SAS disks, which increased the
number of disks of this performance tier by about 10 percent. VMTurbo correctly
analyzed the performance of the storage and migrated disks from several VMs to help
alleviate I/O issues on the most heavily used VM volumes to help reduce overall
congestion on the storage, but did not perform a larger migration to prevent
overloading the performance capabilities of the new disks. As the workload progressed,
the VMTurbo optimizations of the previous day meant that the workload was further
processor-constrained, and began hot-adding vCPUs to compensate.
Later that day, VMTurbo requested more storage resources because of disk
latency, and we provisioned a new RAID 10 pool of ten 10K SAS disks, which doubled the
number of disks of the lowest performance tier. Again, VMTurbo decided to migrate a
small number of VMs when necessary, preventing moving a high-performance workload
to slower storage, while avoiding overloading the new disks (Pools 7 and 8; see Figure
13).
Day 3 brought the first host provisioning decision. We added an identical host to
the cluster to match the first two. VMTurbo decided on inactive VMs to migrate to the
new host one at a time, minimizing the effect of the migrations on the network. This is
in contrast to other solutions, which have a tendency to mass-migrate VMs when a new
resource is presented. While this form of resource distribution can be effective, it may
choke active resources in the environment, causing temporary performance issues.
Day 4 brought another host provisioning decision, and a series of migrations to
balance the workload across hosts. Once the initial migrations were complete, VMTurbo
reassessed the relative load of the four hosts of the cluster, and migrated VMs to
underutilized hosts as the workload progressed through the remainder of the day.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 6
On Day 5, the environment approached equilibrium with the workload. As
Figures 2 through 5 show, the workload optimizations reached a plateau, and VMTurbo
no longer requested further disk pools be allocated to the environment.
Figure 6: VMTurbo decision timeline.
Figure 7: VMTurbo decision timeline.
Throughout the test, the application and fabric control modules also had
decisions for how to improve the efficiency of our environment. Delving into the SQL
performance data, VMTurbo decided that increasing the transaction log size on highusage databases could improve performance, instead of a more costly, brute-force
method of only increasing vCPU or vMemory allocation. The fabric module recognized
that one of the server links in a port channel was underutilized, and suggested removing
it from the group—a move which would not affect performance, but instead save on
expensive per-port fabric licenses.
CONCLUSION
There’s no reason to leave anything on the table when it comes to assuring
application performance at scale. VMTurbo Operations Manager can help you control
resource and storage allocation in a simple, straightforward way and ensure that you’re getting even more out of your databases. We found that adding VMTurbo to a virtual
environment enabled us to better use existing resources, and showed us when
additional hardware was needed to reach peak performance on our workload.
Implementing these decisions increased orders per minute by up to 33 percent and
delivered 23 percent better response times over our week of testing.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 7
For users, VMTurbo can enable better response times, which means less waiting
and frustration. This might mean that a customer can check out faster in an online store
or that an employee can pull up a file instantaneously, both of which mean happier
people.
For administrators, VMTurbo can make day-to-day life easier because it
removes a time-consuming repetitive task from their workday. VMTurbo can find ways
to optimize resources more efficiently, so it takes less time to get better results and
frees administrators up to focus on more important projects.
For organizations, VMTurbo can help cut costs because it enables better
performance from hardware you already have. That way, you don’t have to invest in hardware and expand before you really need to, because you’re maximizing what the resources you have can do.
If you’re looking to assure application performance and get even more out of
your VMware vSphere virtual environment, VMTurbo Operations Manager can help
optimize your environment and make CEOs, administrators, and users happy.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 8
APPENDIX A – ABOUT THE COMPONENTS
About VMTurbo Operations Manager
According to VMTurbo, “VMTurbo’s Application Performance Control system enables customers to guarantee
Quality of Service for any application while maximizing resource utilization of any infrastructure.
“VMTurbo’s patented technology continuously matches any application workload demand to any infrastructure
supply. With this unique, real-time capability, VMTurbo is the only technology capable of controlling and maintaining an
environment in a healthy state.”
Visit the VMTurbo website at www.vmturbo.com to learn more.
About DVD Store Version 2.1
To create our real-world ecommerce workload, we used the DVD Store Version 2.1 benchmarking tool. DS2
models an online DVD store, where customers log in, search for movies, and make purchases. DS2 reports these actions
in orders per minute that the system could handle, to show what kind of performance you could expect for your
customers. The DS2 workload also performs other actions, such as adding new customers, to exercise the wide range of
database functions you would need to run your ecommerce environment.
For more details about the DS2 tool, see www.delltechcenter.com/page/DVD+Store.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 9
APPENDIX B – SYSTEM CONFIGURATION INFORMATION
Figure 8 provides detailed configuration information for the test systems. All four Cisco UCS® B200 M3 blades
under test were configured identically.
System
Enclosure
Blade enclosure
Power Supplies
Total number
Wattage of each (W)
Cooling fans
Total number
General
Number of processor packages
Number of cores per processor
Number of hardware threads per core
System power management policy
CPU
Vendor
Name
Model number
Stepping
Socket type
Core frequency (GHz)
Bus frequency
L1 cache
L2 cache
L3 cache
Platform
Vendor and model number
Motherboard model number
BIOS name and version
BIOS settings
Memory module(s)
Total RAM in system (GB)
Vendor and model number
Type
Speed (MHz)
Speed running in the system (MHz)
Size (GB)
Number of RAM module(s)
Chip organization
VMTurbo enables greater performance for large virtual
environments
Cisco UCS B200 M3 Blade Server
Cisco UCS 5108
4
2,500
8
2
8
2
High Performance
Intel®
Xeon®
E5-2650
6
LGA2011
2.0
8.0 GT/s
512 KB
2 MB
20 MB
Cisco UCS B200 M3
73-14689-04
Cisco® B200M3.2.2.4a.0.041720151158
Default
384
Hynix® UCS-MR-1X162RZ-A
DDR3-1866 PC3-14900
1,866
1,866
16
24
Double-sided
A Principled Technologies report 10
System
Operating system
Name
Build number
Language
RAID controller
Vendor and model number
Firmware version
Network adapter
Vendor and model number
Cisco UCS B200 M3 Blade Server
VMware vSphere 6.0
2143827
English
LSI Logic® MegaRAID SAS 2004 ROMB
20.12.1-0206|4.37.00|NA
Cisco UCS VIC 1240
Figure 8: System configuration information for the test systems.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 11
APPENDIX C – HOW WE TESTED
To test VMTurbo’s ability to improve performance, we created a two-node cluster in VMware vCenter™ 6.0 and evenly distributed an OLTP-based workload across both hosts, and used two nodes for an infrastructure cluster for the
test environment. Additionally, we created two reserve datastores and configured two reserve ESXi hosts to add to the
environment if VMTurbo indicated more raw capacity was required.
We configured two datastores per disk pool on the Fibre Channel storage to simulate a typical environment
involving contention due to multiple datastores per pool.
To simulate a realistic environment of a large database farm, we deployed test clients which requested data
from the Microsoft® SQL servers running several different workloads on three different database sizes. Each Microsoft
SQL Server® was configured with separate OS, transaction log, and database volumes. We used DVD Store as the
benchmark tool to generate the workload and evaluate the performance of the environment. The combined workload
ran for a total of eight hours, with five VMs from each workload segment running for two hours.
During DRS testing, we configured DRS to auto-accept level 1, 2, and 3 suggestions, while Storage DRS was
configured to automatically run, including I/O optimizations at the default levels.
We configured VMTurbo to automatically relocate VMs to hosts or storage, and automatically hot-add vCPU and
memory. All other VMTurbo recommendations were manually implemented during testing.
Test bed configuration
Figure 9 shows a diagram of our test bed. The blade chassis was connected to an 8Gb Fibre Channel storage
array.
Figure 9: Test bed diagram.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 12
Configuring the test bed
Installing VMware ESXi™ 6.0 on the infrastructure servers and servers under test
1. Connect to the UCS server via KVM, and attach the installation media.
2. Boot the server.
3. At the VMware Installer screen, press Enter.
4. At the EULA screen, press F11 to Accept and Continue.
5. Under Storage Devices, select the appropriate virtual disk, and press Enter.
6. For keyboard layout, select US, and press Enter.
7. Enter the root password twice, and press Enter.
8. To start installation, press F11.
9. After the server reboots, press F2, and enter root credentials.
10. Select Configure Management Network, and press Enter.
11. Select the appropriate network adapter, and click OK.
12. Log into the server using the VMware vSphere client.
13. Select the Configuration tab, and click Networking.
14. Click Add Networking…
15. Create the management network on VLAN 102.
16. Click the Configuration tab, and select Time configuration.
17. Click DNS and Routing, and confirm the settings.
Setting up Windows Server® VMs
Create VMs for DC1, Infra-SQL, and vCenter for the infrastructure. The same procedure is used to create the test
VMs. See Figure 11 for configuration values.
1. Log into the ESX server with the VMware vSphere client.
2. In the VMware vSphere client, under Basic Tasks, select Create a new virtual machine.
3. Choose Custom, and click Next.
4. Assign a name to the virtual machine, and click Next.
5. Select the host, and click Next.
6. Select the appropriate storage, and click Next.
7. Choose Virtual Machine Version 11, and click Next.
8. Choose Windows®, select Microsoft Windows Server 2012 R2 (64-bit), and click Next.
9. For CPUs, select the correct number of vCPUs based on the VM role, and click Next.
10. Choose the amount of memory based on the VM role, and click Next.
11. Click 1 for the number of NICs, select VMXNET3, connect to the PRIV-NET port group, and click Next.
12. Leave the default virtual storage controller, and click Next.
13. Keep the default virtual device node (0:0), and click Next.
14. Connect the VM virtual CD-ROM to the Microsoft Windows 2012 R2 installation disk.
15. Click Finish.
16. Right-click the vCenter VM, and click Edit settings.
17. Start the VM.
Installing Microsoft Windows Server 2012 R2
1. Choose the language, time and currency, and keyboard input. Click Next.
2. Click Install Now.
3. Select Windows Server 2012 R2 Datacenter, and click Next.
4. Accept the license terms, and click Next.
5. Click Custom.
6. Click the Disk, and click Next.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 13
7. After the installation completes, enter the administrator password twice, and click Finish.
Configuring the operating system
1. Enable remote desktop access.
2. Change the hostname, and reboot when the installer prompts you.
3. Set up networking:
a. Click Start Control Panel, right-click Network Connections, and select Open.
b. Right-click the management traffic NIC, and select Properties.
c. Select TCP/IP (v4), and select Properties.
d. Set the IP address and subnet for the NIC, which will handle management traffic, and click OK.
e. Click OK, and click Close.
4. Install all available Windows updates. Restart as necessary.
5. Once DC1 is configured as an Active Directory® server, join all other VMs to the AD before continuing.
Installing Active Directory and DNS services
1. Launch Server Manager.
2. Click Manage Add Roles and Features.
3. At the Before you begin screen, click Next.
4. At the Select installation type screen, leave Role-based or feature-based installation selected, and click Next.
5. At the Server Selection Screen, select the server from the pool, and click Next.
6. At the Select Server Roles screen, select Active Directory Domain Services. Click Add Features when prompted,
and click Next.
7. At the Select Features screen, click Next.
8. At the Active Directory Domain Services screen, click Next.
9. At the Confirm installation selections screen, check Restart the destination server automatically if required, and
click Install.
Configuring Active Directory and DNS services
1. After the installation completes, a screen should pop up with configuration options. If not, click the Tasks flag in
the upper-right section of Server Manager.
2. Click Promote this server to a Domain Controller.
3. At the Deployment Configuration screen, select Add a new forest. In the Root domain name field, type
test.local and click Next.
4. At the Domain Controller Options screen, leave the default values, and enter a password twice.
5. Click Next four times to accept default settings for DNS, NetBIOS, and directory paths.
6. At the Review Options screen, click Next.
7. At the Prerequisites Check dialog, allow the check to complete. If there are no relevant errors, check Restart the
destination server automatically if required, and click Install.
8. When the server restarts, log on using test\Administrator and the specified password.
Configuring the Windows Time Service on DC1
To ensure reliable time, we pointed our Active Directory server to a local NTP server.
1. Open a command prompt.
2. Type the following:
W32tm /config /syncfromflags:manual /manualpeerlist:"<ip address of a NTP
server>"
W32tm /config /reliable:yes
W32tm /config /update
W32tm /resync
Net stop w32time
Net start w32time
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 14
Installing and configuring Microsoft SQL Server 2014
1. Open the console for the VM, or connect to the VM with RDP.
2. Mount the installation ISO to the VM.
3. Click to run SETUP.EXE. If Autoplay does not begin the installation, navigate to the SQL Server 2014 DVD, and
double-click.
4. If the installer prompts you with a .NET installation prompt, click Yes to enable the .NET Framework Core role.
5. In the left pane, click Installation.
6. Click New SQL Server stand-alone installation or add features to an existing installation.
7. At the Setup Support Rules screen, wait for the rule check to complete. If there are no failures or relevant
warnings, click OK.
8. Select Specify a free edition, and from the drop-down menu, select Evaluation. Click Next.
9. Click the checkbox to accept the license terms, and click Next.
10. If no failures are displayed after the setup support files are installed, click Next.
11. At the Setup Role screen, choose SQL Server Feature Installation.
12. At the Feature Selection screen, select Database Engine Services, Full-Text and Semantic Extractions for Search,
Client Tools Connectivity, Client Tools Backwards Compatibility, Management Tools – Basic, and Management
Tools – Complete. Click Next.
13. At the Installation Rules screen, after the check completes, click Next.
14. At the Instance configuration screen, leave the default selection of default instance, and click Next.
15. At the Disk Space Requirements screen, click Next.
16. At the Server Configuration screen, check that NT AUTHORITY\SYSTEM is selected for SQL Server Agent and SQL
Server Database Engine. Click Next.
17. Assuming there are no failures, click Next.
18. At the Database Engine Configuration screen, select Mixed Mode.
19. Enter and confirm a password for the system administrator account.
20. Click Add Current user. This may take several seconds.
21. Click the Data Directories tab, and select the default data locations and log locations.
22. Click Next.
23. At the Error and usage reporting screen, click Next.
24. At the Installation Configuration Rules screen, check that there are no failures or relevant warnings, and click
Next.
25. At the Ready to Install screen, click Install.
26. After installation completes, click Close.
27. Download and install any available updates for Microsoft SQL Server 2014.
Setting up an ODBC DSN for vCenter
1. Create new databases on Infra-SQL in Microsoft SQL Management Studio named vcenter and vupdate
2. Click Options, change the recovery model from full to simple on both databases, and click OK.
3. Log into vCenter as TEST\administrator
4. From the desktop of the vCenter server, select Start Run, and type odbcad32.exe. Press Enter.
5. Click the system DSN tab.
6. Click Add.
7. Click SQL Server Native Client 11.0, and click Finish.
8. In the Create a New Data Source to SQL Server text box, type the connection name vcenter
9. For Server, select SQL, and click Next.
10. Change authentication to With SQL Server authentication using a login ID and password entered by the user,
type sa as the Login ID, use the password you defined in SQL server setup for the SA account, and click Next.
11. Select Change the default database to choose vCenter from the pull-down menu, and click Next.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 15
12. Click Finish.
13. To confirm correct configuration, click Test Data Source… 14. To create the vCenter ODBC connection, click OK.
15. Repeat the procedure to create a connection for the vupdate database.
Installing VMware vCenter 6.0 on Windows Server 2012 R2
1. From the VMware vCenter 6.0 install media, double-click Autorun.exe.
2. To start the install wizard, click Install.
3. Click Next.
4. Check the box to accept the license agreement, and click Next.
5. Leave the default radio button selection (vCenter Server and Embedded Platform Services Controller), and click
Next.
6. Provide the correct system FQDN, and click Next.
7. Select the radio button to create a new Single Sign-on domain, and fill in all necessary information. Click Next.
8. Choose to run the vCenter service under a domain account or the Windows Local System Account. If a domain
account, provide the username and password. Click Next.
9. Provide database credentials to use the database created previously. Click Next.
10. Leave all ports configured to their default, and click Next.
11. Choose a location for vCenter to install files, and choose a directory for vCenter data. Click Next.
12. Verify all settings on the Ready to install screen, and click Install.
13. After install has completed, select Finish.
14. On the installer, click Exit.
15. Restart the server.
16. Navigate to the IP address of the vCenter installation in a web browser, and log into the vCenter server using the
domain account TEST\administrator
17. Right-click the root of vCenter, and click New Data center.
18. Name the new datacenter datacenter
19. Add all ESX servers under test to the datacenter.
Installing Internet Information Services
1. Click Start Administrative Tools Server Manager.
2. On the left pane, click Roles.
3. Click Add Roles.
4. Click the Application Server checkbox.
5. When the Add features required for Web Server (IIS)? screen appears, click Add Features.
6. Select .NET Framework 3.5 and 4.5, and add ASP.NET 4.5.
7. Click Next.
8. Click Next.
9. At the Select Role Services page for Web Server (IIS), click Next.
10. Click Install.
11. Click Close.
Configuring Internet Information Services
1. Copy the pre-configured ds2web folder containing the HTML and .cs files for the DS2 Web application to C:\
2. Open Internet Information Services (IIS) Manager.
3. Expand the server tree, and disable the Default Web Site.
4. Add a website named ds2
5. Set the Physical path to C:\ds2web
6. Click OK.
7. Add a Virtual Directory, and point the virtual path /ds2 to the root application.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 16
Configuring the database workload client
For our testing, we used a virtual client for the IIS/Microsoft SQL Server client. To create this client, we installed
Windows Server 2012 R2, assigned a static IP address, and installed .NET 4.5.
Configuring the database
Data generation overview
We generated the data using the Install.pl script included with DVD Store version 2.1 (DS2), providing the
parameters for our 5GB, 20GB, and 40GB database sizes and the Microsoft SQL Server 2014 platform. We ran the
Install.pl script on a utility system running Linux®. The Install.pl script also generated the database schema.
After processing the data generation, we transferred the data files and schema creation files to a Windowsbased system running SQL Server 2012. We built the databases in SQL Server 2012, and then performed a full backup,
storing the backup file on a network share for quick access. We used that backup file to restore the servers between test
runs.
The only modification we made to the schema creation scripts were the specified file sizes for our databases.
We explicitly set the file sizes higher than necessary to ensure that no file-growth activity would affect the outputs of the
test. Besides this file size modification, the database schema was created and loaded according to the DVD Store
documentation. Specifically, we followed the steps below:
1. We generated the data and created the database and file structure using database creation scripts in the DS2
download. We made size modifications specific to our databases and the appropriate changes to drive letters.
2. We transferred the files from our Linux data generation system to a Windows system running SQL Server.
3. We created database tables, stored procedures, and objects using the provided DVD Store scripts.
4. We set the database recovery model to bulk-logged to prevent excess logging.
5. We loaded the data we generated into the database. For data loading, we used the import wizard in SQL Server
Management Studio. Where necessary, we retained options from the original scripts, such as Enable Identity
Insert.
6. We created indices, full-text catalogs, primary keys, and foreign keys using the database-creation scripts.
7. We updated statistics on each table according to database-creation scripts, which sample 18 percent of the
table data.
8. On the SQL Server instance, we created a ds2user SQL Server login using the following Transact-SQL (T-SQL)
script:
USE [master]
GO
CREATE LOGIN [ds2user] WITH PASSWORD=N’’,
DEFAULT_DATABASE=[master],
DEFAULT_LANGUAGE=[us_english],
CHECK_EXPIRATION=OFF,
CHECK_POLICY=OFF
GO
9. We set the database recovery model back to full.
10. We created the necessary full text index using SQL Server Management Studio.
11. We created a database user and mapped this user to the SQL Server login.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 17
12. We then performed a full backup of the database. This backup allowed us to restore the databases to a pristine
state relatively quickly between tests.
Figures 10 through 12 show our initial file size modifications.
Logical name
Database files (5GB database)
Primary
cust1
cust2
ind1
ind2
ds_misc
orders1
orders2
Log files
ds_log
Filegroup
Initial size (MB)
PRIMARY
DS_CUST_FG
DS_CUST_FG
DS_IND_FG
DS_IND_FG
DS_MISC_FG
DS_ORDERS
DS_ORDERS
4
2,560
2,560
1,536
1,536
1,536
1,536
1,536
Not applicable
10,240
Figure 10: Our initial file size modifications.
Logical name
Database files (20GB database)
Primary
cust1
cust2
cust3
cust4
ind1
ind2
ind3
ind4
ds_misc
orders1
orders2
orders3
orders4
Log files
ds_log
Filegroup
Initial size (MB)
PRIMARY
DS_CUST_FG
DS_CUST_FG
DS_CUST_FG
DS_CUST_FG
DS_IND_FG
DS_IND_FG
DS_IND_FG
DS_IND_FG
DS_MISC_FG
DS_ORDERS
DS_ORDERS
DS_ORDERS
DS_ORDERS
5
6,144
6,144
6,144
6,144
2,891
2,891
2,891
2,891
200
3,072
3,072
3,072
3,072
Not applicable
18,141
Figure 11: Our initial file size modifications.
Logical name
Database files (40GB database)
Primary
cust1
cust2
cust3
cust4
Filegroup
PRIMARY
DS_CUST_FG
DS_CUST_FG
DS_CUST_FG
DS_CUST_FG
VMTurbo enables greater performance for large virtual
environments
Initial size (MB)
5
12,288
12,288
12,288
12,288
A Principled Technologies report 18
Logical name
ind1
ind2
ind3
ind4
ds_misc
orders1
orders2
orders3
orders4
Log files
ds_log
Filegroup
DS_IND_FG
DS_IND_FG
DS_IND_FG
DS_IND_FG
DS_MISC_FG
DS_ORDERS
DS_ORDERS
DS_ORDERS
DS_ORDERS
Initial size (MB)
5,826
5,826
5,826
5,826
269
6,144
6,144
6,144
6,144
Not applicable
35,437
Figure 12: Our initial file size modifications.
Configuring storage
Figure 13 provides information on the configuration of the storage for the vSphere clusters and the mount
points.
Storage, volumes, and storage pools
Enclosure
Storage pool 1
RAID level
RAID 5 (8+1)
200
Drive size (GB)
Drive RPM or SSD
SSD
Number of drives
9
Drive type
6Gb SAS 3.5ʺ″
Storage pool 2
RAID level
RAID 10
300
Drive size (GB)
Drive RPM or SSD
15K
Number of drives
16
Drive type
6Gb SAS 3.5ʺ″
Storage pool 3
RAID level
RAID 5 (4+1)
600
Drive size (GB)
Drive RPM or SSD
10K
Number of drives
10
Drive type
6Gb SAS 3.5ʺ″
Storage pool 4
RAID level
RAID 10
300
Drive size (GB)
Drive RPM or SSD
15K
Number of drives
16
Drive type
6Gb SAS 3.5ʺ″
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 19
Storage, volumes, and storage pools
Storage pool 5
RAID level
Drive size (GB)
Drive RPM or SSD
Number of drives
Drive type
Storage pool 6
RAID level
Drive size (GB)
Drive RPM or SSD
Number of drives
Drive type
Storage pool 7
RAID level
Drive size (GB)
Drive RPM or SSD
Number of drives
Drive type
Storage pool 8
RAID level
Drive size (GB)
Drive RPM or SSD
Number of drives
Drive type
RAID 10
300
15K
24
6Gb SAS 3.5ʺ″
RAID 10
300
15K
32
6Gb SAS 3.5ʺ″
RAID 10
300
15K
8
6Gb SAS 3.5ʺ″
RAID 10
600
10K
10
6Gb SAS 3.5ʺ″
Figure 13: Configuration information for the Generic NAS solution.
Configuring VMTurbo
Installing VMTurbo
1. Install the appliance OVF using vCenter.
a. Select File Deploy OVF Template.
b. Browse to the location of the OVA file.
c. Continue through the wizard, and accept the license agreement.
d. Name the new appliance VM.
e. Select the destination datastore, and choose the provision method.
f. Map the VM network to the test network.
g. Select Power on after deployment, and click Finish.
2. Open a console to the VMTurbo appliance.
3. Log into the appliance with the default ipsetup/ipsetup credentials.
4. Set the static IP address for the appliance.
5. Confirm the change.
6. Log in as root/vmturbo
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 20
7.
8.
9.
10.
11.
12.
13.
14.
Execute yast timezone and set the correct time zone and system time.
Connect to the IP address via web browser.
Log in as administrator/administrator
Install the VMTurbo license.
Navigate to Admin User Configuration.
Enter the Active Directory Domain and Server information, and click Apply.
Navigate to Admin Maintenance.
Select Check for software updates, and apply any available updates.
Configuring Admin
1. Navigate to Admin Target Configuration.
2. Select Add, and select Hypervisor.
3. Enter the host name/IP address and authentication credentials for the vCenter, and click Apply.
4. Select Add, and select Storage.
5. Enter the IP address of the SMI-S provider and authentication credentials for the block level storage, and click
Apply.
6. Select Add, and select Fabric.
7. Enter the IP address of the Cisco UCS chassis and authentication credentials, and click Apply.
8. Select Add, and select Database.
9. Select the scope of SQL VMs under test and enter authentication credentials, and click Apply.
10. Allow at least 15 minutes for the VMTurbo appliance to discover the infrastructure.
Configuring Policy
1. Under Action VM Virtual Machines by PM cluster, enable the overrides for Move, Storage Move, and Resize
up, and set the value to Automated.
2. Under Discovery Application Discovery Application Credentials Virtual Machines, override the Username
and Password attributes, and enter the administrator authentication credentials.
3. Under Analysis, check the box Ignore Hyperthreading.
4. Under Analysis Utilization Constraints, change the Storage Amount and IOPS thresholds to 80 percent, Host
CPU Utilization to 60 percent, and Host Memory to 80 percent.
Test start and run times
We ran the workloads in staggered groups in order to start and record all performance counters for this report.
The specific testing parameters we used for each benchmark are included in the setup sections, and we describe
specifics for launching these tests below.
About running the DVD Store tests
We created a series of batch files, SQL scripts, and shell scripts to automate the complete test cycle. DVD Store
outputs an orders-per-minute metric, which is a running average calculated through the test. In this report, we report
the last OPM reported by each client/target pair.
Each complete test cycle consisted of the general steps listed below.
1.
2.
3.
4.
5.
Reboot the ESX hosts.
Clean up prior outputs from the target system and the client driver system.
Drop the database from the target.
Restore the database on the target.
Wait for a ping response from the server under test (the hypervisor system), the client system, and the target.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 21
6. Let the test server idle for 5 minutes.
7. Start the DVD Store driver on the client.
8. We used the following DVD Store parameters for testing (5GB database example):
ds2sqlserverdriver.exe --target=<target_IP> --ramp_rate=100 --run_time=120 -n_threads=8 --db_size=5GB --think_time=0.003 --detailed_view=Y -warmup_time=1 --report_rate=1 --csv_output=<drive path>
9. Monitor the VMTurbo console for recommended changes and make note of recommendations.
10. Once the 60-minute testing cycle completes, collect the performance data and reset the environment, per step 2
above.
11. Implement all recommendations given by VMTurbo during the first test.
12. Begin a second test run, repeating steps 3 through 8, this time implementing any non-disruptive VMTurbo
recommendations during the test run (e.g., host migrations).
13. Collect the performance data, implement any disruptive recommendations, and reset the environment.
14. Perform a third test run, implement non-disruptive VMTurbo recommendations, and collect the performance
data.
1-20
21-40
41-60
61-80
81-100
DB_SIZE
5GB
40GB
20GB
40GB
40GB
THREAD_COUNT
8
24
12
32
12
RUN_TIME
120
120
120
120
120
THINK_TIME
0.003
0.003
0.003
0.003
0.003
WARMUP_TIME
1
1
1
1
1
RAMP_RATE
100
100
100
100
100
Figure 14: DS2 configuration groups.
Detailed virtual machine configuration
Figures 15 and 16 show the detailed information of the virtual machines we used in testing before implementing
VMTurbo recommendations. vDisk sizes are Thin Provisioned allocation.
VM name
Qty
SQL Server 2014 (Small)
SQL Server 2014 (Large)
40
60
OS
Mem
(GB)
vCPUs
Windows 2012 R2 2
Windows 2012 R2 2
4
4
OS vDisk DB vDisk
(GB)
(GB)
40
60
40
120
Log vDisk
(GB)
80
40
Figure 15: Detailed information for our virtual machines.
VM storage location groups
GB
VM
5
1-20 (Odd)
5
1-20 (Even)
40
21-38 (Odd)
40
21-38 (Even)
40
39
40
40
20
41-58 (Odd)
20
41-58 (Even)
Load
Light
Light
Med-High
Med-High
Med-High
Med-High
Medium
Medium
VMTurbo enables greater performance for large virtual
environments
Datastore
Pool3-A
Pool3-B
Pool5-A
Pool5-B
Pool6-A
Pool6-B
Pool2-A
Pool2-B
LUN
2
8
4
10
Capacity (GB)
3,200
3,200
2,370
2,370
1
7
1,575
1,575
A Principled Technologies report 22
VM storage location groups
GB
VM
20
59
20
60
40
61,66,71
40
62,67,76
40
61-80 (Odd)
40
61-80 (Even)
40
81-92 (Odd)
40
81-92 (Even)
40
93-100 (Odd)
40
93-100 (Even)
Load
Medium
Medium
High
High
High
High
Medium
Medium
Medium
Medium
Datastore
Pool3-A
Pool3-B
Pool1-A
Pool1-B
Pool6-A
Pool6-B
Pool4-A
Pool4-B
Pool6-A
Pool6-B
LUN
Capacity (GB)
0
6
5
11
3
9
948
948
3,162
3,162
1,575
1,575
Figure 16: VM storage location groups.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 23
ABOUT PRINCIPLED TECHNOLOGIES
Principled Technologies, Inc.
1007 Slater Road, Suite 300
Durham, NC, 27703
www.principledtechnologies.com
We provide industry-leading technology assessment and fact-based
marketing services. We bring to every assignment extensive experience
with and expertise in all aspects of technology testing and analysis, from
researching new technologies, to developing new methodologies, to
testing with existing and new tools.
When the assessment is complete, we know how to present the results to
a broad range of target audiences. We provide our clients with the
materials they need, from market-focused data to use in their own
collateral to custom sales aids, such as test reports, performance
assessments, and white papers. Every document reflects the results of
our trusted independent analysis.
We provide customized services that focus on our clients’ individual requirements. Whether the technology involves hardware, software,
websites, or services, we offer the experience, expertise, and tools to
help our clients assess how it will fare against its competition, its
performance, its market readiness, and its quality and reliability.
Our founders, Mark L. Van Name and Bill Catchings, have worked
together in technology assessment for over 20 years. As journalists, they
published over a thousand articles on a wide array of technology subjects.
They created and led the Ziff-Davis Benchmark Operation, which
developed such industry-standard benchmarks as Ziff Davis Media’s Winstone and WebBench. They founded and led eTesting Labs, and after
the acquisition of that company by Lionbridge Technologies were the
head and CTO of VeriTest.
Principled Technologies is a registered trademark of Principled Technologies, Inc.
All other product names are the trademarks of their respective owners.
Disclaimer of Warranties; Limitation of Liability:
PRINCIPLED TECHNOLOGIES, INC. HAS MADE REASONABLE EFFORTS TO ENSURE THE ACCURACY AND VALIDITY OF ITS TESTING, HOWEVER,
PRINCIPLED TECHNOLOGIES, INC. SPECIFICALLY DISCLAIMS ANY WARRANTY, EXPRESSED OR IMPLIED, RELATING TO THE TEST RESULTS AND
ANALYSIS, THEIR ACCURACY, COMPLETENESS OR QUALITY, INCLUDING ANY IMPLIED WARRANTY OF FITNESS FOR ANY PARTICULAR PURPOSE.
ALL PERSONS OR ENTITIES RELYING ON THE RESULTS OF ANY TESTING DO SO AT THEIR OWN RISK, AND AGREE THAT PRINCIPLED
TECHNOLOGIES, INC., ITS EMPLOYEES AND ITS SUBCONTRACTORS SHALL HAVE NO LIABILITY WHATSOEVER FROM ANY CLAIM OF LOSS OR
DAMAGE ON ACCOUNT OF ANY ALLEGED ERROR OR DEFECT IN ANY TESTING PROCEDURE OR RESULT.
IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES, INC. BE LIABLE FOR INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES IN
CONNECTION WITH ITS TESTING, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES,
INC.’S LIABILITY, INCLUDING FOR DIRECT DAMAGES, EXCEED THE AMOUNTS PAID IN CONNECTION WITH PRINCIPLED TECHNOLOGIES, INC.’S TESTING. CUSTOMER’S SOLE AND EXCLUSIVE REMEDIES ARE AS SET FORTH HEREIN.
VMTurbo enables greater performance for large virtual
environments
A Principled Technologies report 24
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising