DELL Force10 S6000 as a physical switch for VMware NSX

Based on this  document http://www.vmware.com/files/pdf/products/nsx/vmw-nsx-dell-systems.pdf
DELL Force10 S6000 is going to be fully integrated with VMware NSX (NSX is software defined networking platform).

Dell Networking provides:

• Data center switches for robust underlays for L2 overlays
• CLI for virtual and physical networks
• Network management and automation with Active Fabric Manager
• S6000 Data Center Switch Gateway for physical workloads to connect to virtual networks
• Complete end-to-end solutions that include server, storage, network, security, management and services with world wide support

Dell S6000 use cases:

• Extend virtual networks to physical servers -  S6000 works as VXLAN gateway to VLANs on physical network (VXLAN VTEP).
• Connect physical workloads reachable on a specific VLAN to logical networks via an L2 service
• Connect physical workloads reachable on a specific port to logical networks via an L2 service
• Connect to physical workloads in a Physical to virtual migration
• Migration from existing virtualized environments to public clouds, creating hybrid clouds
• Access physical router, firewall, load balancer, WAN optimization and other network resources

I cannot wait to test it in my lab or on customer PoC engagement. After hands-on experience I'll share it on this blog. 

What’s New in vSphere 5.5

On this article I'll try to collect all important (at least for me) vSphere 5.5 news and improvements announced at VMworld 2013. I wasn't there so I rely on other blog posts and VMware materials.

Julian Wood reported about vCloud Suite 5.5 news announced at VMworld 2013 at
http://www.wooditwork.com/2013/08/26/whats-new-vcloud-suite-5-5-introduction/

Chris Wahl wrote deep dive blog posts into vSphere 5.5 improvements at
http://wahlnetwork.com/category/deep-dives/5-5-vsphere-improvements/

Cormac Hogan listed storage improvements in vSphere 5.5 at
http://blogs.vmware.com/vsphere/2013/08/whats-new-in-vsphere-5-5-storage.html?utm_source=twitterfeed&utm_medium=linkedin&goback=%2Egde_3217230_member_269944857#%21 

Thanks Julian, Chris, and Cormac for excellent blog posts and keep informed as who was not able to attend VMworld 2013.

BTW: Official VMware What's New paper is at http://www.vmware.com/files/pdf/vsphere/VMware-vSphere-Platform-Whats-New.pdf 

Here are few citations with my comments from above blog posts. I'll mention just improvements which are important and/or interesting for me. I will concentrate on these topics and in near future I have to find and test more hidden details.

1. Management: VMware is strongly recommending using a single VM for all vCenter Server core components (SSO, Web Client, Inventory Service and vCenter Server) or to use the appliance rather than splitting things out which just add complexity and makes it harder to upgrade in the future. << "This is excellent approach and I really like it."
    
2. Management: The vCenter Appliance has also been beefed up and with its embedded database supports 300 hosts and 3000 VMs or if you use an external Oracle DB the supported hosts and VMs are the same as for Windows. << "Finally"
    
3. Storage: vSphere 5.5 now supports VMDK disks larger than 2TB. Disks can be created up to 63.36TB in size on both VMFS and NFS.The max disk size needs to be about 1% less than the datastore file size limit. << "The last vSphere storage limit disappeared however how big datastores we will create?"
    
4. Storage: vSphere Flash Read Cache leveraging local SSDs to eliminate read IO operations from datastores and save storage performance (IOPSes) for other purposes (writes, other workloads, etc.)  For more info look at http://wahlnetwork.com/2013/08/26/vsphere-5-5-improvements-part-5-vsphere-flash-read-cache-vflash/ or http://www.yellow-bricks.com/2013/08/26/introduction-to-vsphere-flash-read-cache-aka-vflash/ << "Sounds good but pernixdata.com looks better."
    
5. Storage: vSphere vSAN leveraging SSD and SATA server internal disks and form it into shared storage pool. VMware promised it is match better than VSA (VMware Storage Appliance). For more info look at http://wahlnetwork.com/2013/08/26/vsphere-5-5-improvements-part-4-virtual-san-vsan/ << "We will see. have you tested VSA? I still believe real storage is real storage. At least now. However if someone considers vSAN I would recommend to invest into really good server disks and SSDs."
    
6. Storage: PDL AutoRemove in vSphere 5.5 automatically removes a device with PDL from the host. PDL stands for Permanent Device Lost and receive it from storage array as a SCSI Sense Code. << "It would be beneficial when some storage admin removes empty LUN. Then nothing should be done on vSphere in case storage send appropriate SCSI Sense Code. MUST BE CAREFULLY TEST IT!!!"
    
7. Networking: LACP in 5.5 gives you over 22 load balancing algorithms and you are now able to create 32 LAGs per host so you can bond together all those physical Nics. << "Finally, Nexus 1000v had it already from the beginning."
    
8. Networking: Flow based marking and filtering provides granular traffic marking and filtering capabilities from a simple UI integrated with VDS UI. You can provide stateless filtering to secure or control VM or Hypervisor traffic. Any traffic that requires specific QoS treatment on physical networks can now be granularly marked with COS and DSCP marking at the vNIC or Port group level. << "Nice improvement, but I have never had such requirement so far."
    
9. High Availability: Someone mentioned to me that VMware announced vSphere 5.5  Multi-processor Fault Tolerance (FT) in VMworld 2013.  << "This would be interesting but must be validated as I cannot find any official statement or some blog post about it. It seems to me it was Fault Tolerance tech preview like in VMworld 2012 session I attended last year."
    
10. Authentication:  SSO 2.0 is now a multi-master model. Replication between SSO servers is automatic and built-in. SSO is now site aware. The SSO database is completely removed. For more info look at http://wahlnetwork.com/2013/08/26/vsphere-5-5-improvements-part-7-single-sign-on-completely-redesigned/  << "Finally, previous SSO 1.0 was a nightmare!!!"
     
11. Disaster Recovery: VMware Replication (VR) now supports more VR Server Appliances responsible for replication, more point in time instances (aka snapshots), the ability to use Storage vMotion on protected VMs, and vSphere Web Client will show you details on your vSphere Replication status when you click on the vCenter object. For more info look at http://wahlnetwork.com/2013/08/26/vsphere-5-5-improvements-part-6-site-recovery-manager-srm-and-vsphere-replication/ << "Cool. Good evolution."

DELL OpenManage Essentials (OME)

OpenManage Essentials (OME) is a systems management console that provides simple, basic Dell hardware management and is available as a free download.

DELL OME can be downloaded at https://marketing.dell.com/dtc/ome-software?dgc=SM&cid=259733&lid=4682968

Patch 1.2.1 downloadable at
http://www.dell.com/support/drivers/us/en/555/DriverDetails?driverId=P1D4C

For more information look at DELL Tech Center.

Data Center Bridging

DCB 4 key protocols:

•  Priority-based Flow Control (PFC): IEEE 802.1Qbb
•  Enhanced Transmission Selection (ETS): IEEE 802.1Qaz
•  Congestion Notification (CN or QCN): IEEE 802.1Qau
•  Data Center Bridging Capabilities Exchange Protocol (DCBx)

PFC - provides a link level flow control mechanism that can be controlled independently for each frame priority. The goal of this mechanism is to ensure zero loss under congestion in DCB networks.PFC is independent traffic priority pausing and enablement of lossless packet buffers/queuing for particular 802.1p CoS.

ETS - provides a common management framework for assignment of bandwidth to frame priorities. Bandwidth can be dynamic based on congestion and relative ratios between defined flows. ETS provides minimum, guaranteed bandwidth allocation per traffic class/priority group during congestion and permits additional bandwidth allocation during non-congestion.

CN - provides end to end congestion management for protocols that are capable of transmission rate limiting to avoid frame loss. It is expected to benefit protocols such as TCP that do have native congestion management as it reacts to congestion in a more timely manner. Excellent blog post about CN is here.

DCBX - a discovery and capability exchange protocol that is used for conveying capabilities and configuration of the above features between neighbors to ensure consistent configuration across the network. Performs discovery, configuration, and mismatch resolution using Link Layer Discovery Protocol (IEEE 802.1AB - LLDP).

DCBX can be leveraged for many applications.
One DCBX application example is iSCSI application priority - Support for the iSCSI protocol in the application priority DCBX Type Length Value (TLV). Advertises the priority value (IEEE 802.1p CoS, PCP field in VLAN tag) for iSCSI protocol. End devices identify and tag Ethernet frames containing iSCSI data with this priority value.

DELL Force10 I/O Aggregator 40Gb Port Question

Today I have received question how to inter connect DELL Force10 IOA 40Gb uplink with DELL Force10 S4810 top of rack switches.

I assume the reader is familiar with DELL Force10 datacenter networking portfolio.

Even if you have 40Gb<->40Gb twinax cable with QSFPs between IOA and Force10 S4810 switch it is in IOA side configured by default as 4x10Gb links grouped  in Port-Channel 128.

If you connect it directly into 40Gb port in Force10 S4810 switch the 40Gb port is by default configured as 1x40Gb interface.

That’s the reason why it doesn’t work out-of-the-box. Port speeds are simply mismatched.

To make it correct you have to change 40Gb switch port to 4x10Gb port. Here is the S4810 command to change switch port from 1x40Gb to 4x10Gb:

stack-unit 0 port 48 portmode quad

Here is the snip from S4810 configuration where 40Gb port 0/48 is configure as 4x10Gb port in port-channel 128

interface TenGigabitEthernet 0/48
no ip address
!
port-channel-protocol LACP
  port-channel 128 mode active
no shutdown
!
interface TenGigabitEthernet 0/49
no ip address
!
port-channel-protocol LACP
  port-channel 128 mode active
no shutdown
!
interface TenGigabitEthernet 0/50
no ip address
!
port-channel-protocol LACP
  port-channel 128 mode active
no shutdown
!
interface TenGigabitEthernet 0/51
no ip address
!
port-channel-protocol LACP
  port-channel 128 mode active
no shutdown

interface Port-channel 128
no ip address
portmode hybrid
switchport
no shutdown

Best Practices for Faster vSphere SDK Scripts

Reuben Stump published excellent blog post at http://www.virtuin.com/2012/11/best-practices-for-faster-vsphere-sdk.html about performance optimization of PERL SDK Scripts.

The main takeaway is to minimize the ManagedEntity's Property Set.

So instead of

my $vm_views = Vim::find_entity_views(view_type => "VirtualMachine") ||

  die "Failed to get VirtualMachines: $!";

you have to use

# Fetch all VirtualMachines from SDK, limiting the property set

my $vm_views = Vim::find_entity_views(view_type => "VirtualMachine",

          properties => ['name', 'runtime.host', 'datastore']) ||

  die "Failed to get VirtualMachines: $!";

This small improvement have significant impact on performance because it eliminates big data (SOAP/XML) generation and transfer between vCenter service and the SDK script.

It helped me improve performance of my script from 25 seconds to just 1 second. And the impact is even better for bigger vSphere environment. So my old version of script was almost useless and this simple improvement help me so much.

Thanks Reuben for sharing this information.
 

DELL Blade Chassis power consumption analytics in vCenter Log Insight

DELL Blade Chassis has a capability to send power consumption information via syslog messages. I have never understood  how to practically leverage this capability. When VMware released vCenter Log Insight I have immediately realized how to leverage this tool to visualize blade chassis power consumption.

I prepared short video how to create blade chassis power consumption graph in vCenter Log Insight. The video is located at http://youtu.be/fda1cLW8enA

ESXi Advanced Settings for NetApp NFS

Here are NetApp 
Net.TcpipHeapSize=30
Net.TcpipHeapMax=120
NFS.MaxVolumes=64
NFS.HeartbeatMaxFailures=10
NFS.HeartbeatFrequency=12
NFS.HeartbeatTimeout=5

Enabled SIOC or if you don't have Entrprise+ license set NFS.MaxQueueDepth=64, 32 or 16 based on storage workload and utilization

Unified Network, DCB and iSCSI challenges

iSCSI SAN is Storage Area Network. Storage need lost less fabric. If, for any reason, unified fabric need to be used then quality of ethernet/IP network  is crucial for problem less storage operation.

For example DELL EqualLogic supports and leverage DCB (PFC, ETS and DCBX).
iSCSI-TLV is a part of DCBX. However the DCB protocol primitives must be supported end to end so if one member of the chain doesn't support it than it is useless.

How DCB makes iSCSI better is deeply explained here. 

So think twice if you really want converged network (aka unified fabric) or dedicated iSCSI network is better option for you.

DELL EqualLogic general recommendations for VMware vSphere ESXi

Bellow are eight major recommendations for DELL EqualLogic implementation with vSphere ESXi:

1. Delayed ACK disabled
2. LRO disabled
3. If using Round Robin, set IOs to 3
4. If  ENTERPRISE or ENTERPRISE+ license and ONLY with Enterprise/Enterprise+  install MEM 1.1.2
5. Extend login_timeout to 60 seconds.
6. Don’t have multiple VMDKs on a single Virtual SCSI controller.  (major cause of latency alerts)
7. Align partitions on 64K boundary
8. Format with 64K cluster (allocation unit) size with Windows

CISCO Nexus 1000v - Quality Of Service configuration

class-map type queuing match-any n1kv_control_packet_mgmt_class
 match protocol n1k_control
 match protocol n1k_packet
 match protocol n1k_mgmt

class-map type queuing match-all vmotion_class
 match protocol vmw_vmotion

class-map type queuing match-all vmw_mgmt_class
 match protocol vmw_mgmt

class-map type queuing match-any vm_production
 match cos 0

policy-map type queuing uplink_queue_policy
 class type queuing n1kv_control_packet_mgmt_class
   bandwidth percent 10
 class type queuing vmotion_class
   bandwidth percent 30
 class type queuing vmw_mgmt_class
   bandwidth percent 10
 class type queuing vm_production
   bandwidth percent 40

port-profile type ethernet uplink
 service-policy type queuing output uplink_queue_policy

How to downgrade IBM V7000 (Storwize) firmware

Sometimes, especially when you do a problem management, you have a need to downgrade firmwares on some system components. I have such need for IBM V7000 storage array. Downgrade process is not documented in IBM official documentation so here is the downgrade process step by step:

1. Double check you have IP addresses on management interfaces of both canisters (controllers)
2. Login to management interface of one particular canister over https. https://[ip_mgmt_canister]. You have to use superuser credentials. Default IBM Storwize superuser password is passw0rd
3. Switch node to serrvice state. You should wait 15-20 minutes
4. Login to second node management interface. 
5. Switch second node to service state. You should wait another 15-20 minutes
6. Double check both nodes are in service state
7. Login to one node and choose action "Reinstall software". Browse and upload firmware image via web browser. Software reinstallation takes a while. You have to wait approximattely one or two hours. In the mean time you can ping canyster management IP addresses to check when nodes comming back. 
8. Repeat software reinstallation for second node.
9. Please be aware that storage configuration is lost after software reinstallation. Therefore you have to use default password for superuser. Recall it is passw0rd
10. When both nodes are up and running login to one canister node management interface and exit both nodes from service state. It can takes another 15-20 minutes.
11. When nodes are active you have to regenerate Cluster ID. You have to go to "Configure Enclosure" and enable checkbox "Reset System ID".
12. After all these actions you have Storwize ready to form a new Cluster. So create cluster and assign cluster virtual IP address you will use for standard storage management.

ESX host remote syslog configuration

For remote CLI you can use vMA or vCLI. Here is the example how to configure ESX host (10.10.1.71) to send logs remotely to syslog server listening on IP address 10.10.4.72 on tcp port 514. 

First of all we have to instruct ESX where is the syslog server.

esxcli -s 10.10.1.71 -u root -p Passw0rd. system syslog config set --loghost='tcp://10.10.4.72:514'

Then syslog service on ESX host must be restarted to accept configuration change.

esxcli -s 10.10.1.71 -u root -p Passw0rd. system syslog reload

ESX firewall must be reconfigured to allow syslog traffic

esxcli -s 10.10.1.71 -u root -p Passw0rd. network firewall ruleset set --ruleset-id=syslog --enabled=true
esxcli -s 10.10.1.71 -u root -p Passw0rd. network firewall refresh

If you want to test or troubleshoot syslog logging you can login to ESX host and use logger command to send test message to syslog.

logger "Test syslog over network"

Excellent article: "Anatomy of an Ethernet Frame"

Trey Layton (aka EthernetStorageGuy) wrote excellent article about MTU sizes and Jumbo Frame settings. The article is here. In the article you will learn what MTU size parameters you have to configure in the path among server, network gear and storage. It is crucial to understand difference between payload (usually 1500 or 9000) and different frame sizes (usually 1522 or 9018 or 9022 or  9216) on networking equipment.

Here is summation of  Trey's deep Ethernet frame anatomy in to the simple best practice. "If you want to implement Jumbo Frames use pure datacenter networking equipment and setup MTU size to the device maximum which is usually 9216."

DELL Open Manage Essentials 1.2 has been released

Dell OpenManage Essentials is a 'one to many' console used to monitor Dell Enterprise hardware. It can discover, inventory, and monitor the health of Dell Servers, Storage, and network devices. Essentials can also update the drivers and BIOS of your Dell PowerEdge Servers and allow you to run remote tasks. OME can increase system uptime, automate repetitive tasks, and prevent interruption in critical business operations.

It can be downloaded here.

Fixes & Enhancements
Fixes:

1. Multiple defect fixes and performance improvements

Enhancements:

1. Support for Discovery, Inventory and Map View for Dell PowerEdge VRTX devices. 
2. Addition of Microsoft Windows Server 2012 as a supported operating system for the management station.
3. Context sensitive Search functionality. 
4. Ability to configure OpenManage Essentials to send the warranty status of your devices through email at periodic intervals. 
5. Ability to configure OpenManage Essentials to generate a warranty scoreboard based on your preference and display a notification icon in the heading banner when the warranty scoreboard is available.
6. Enhanced support for Dell Compellent, Dell Force10 E-Series and C-Series, Dell PowerConnect 8100 series, Dell PowerVault FS7500, and PowerVault NX3500 devices. 
7. Support for installing OpenManage Essentials on the domain controller.
8. Device Group Permissions portal. 
9. Additional reports: Asset Acquisition Information, Asset Maintenance Information, Asset Support Information, and Licensing Information. 
10. Addition of a device group for Citrix XenServers and Dell PowerEdge C servers in the device tree. 
11. Availability of storage and controller information in the device inventory for the following client systems: Dell OptiPlex, Dell Latitude, and Dell Precision.
12. CLI support for discovery, inventory, status polling, and removal of devices from the device tree. 
13. Availability of sample command line remote tasks for uninstalling OpenManage Server Administrator and applying a server configuration on multiple managed nodes. 
14. Support for SUDO users in Linux for system updates and OMSA deploy tasks. 
15. Display of a notification icon in the heading banner to indicate the availability of a newer version of OpenManage Essentials. 
16. Support for enabling or disabling rebooting after system update for out-of band (iDRAC) system updates.
17. Support for re-running system update and OpenManage Server Administrator (OMSA) deployment tasks.
18. Support for Single Sign-On (SSO) for iDRAC and CMC devices. 
19. Ability to log on as a different user.

How to change default path selection policy for particular storage array?

Sometimes the firmware in storage array has some problems and you have to "downgrade" functionality to achieve operable system. That's sometimes happen for some ALUA storage systems where Round Robin path policy or Fixed path policy (aka FIXED) should work but doesn't because of firmware issue.

So relatively simple solution is to switch back from more advanced round robin policy to legacy - but properly functioning -  Most Recently Used path policy (aka MRU) normally used for active/passive storage systems. 

Note: Please be aware that some storage vendors saying they have active/active storage even they have not. Usually and probably more precisely they call it "dual-active storage" which is not same as active/active. Maybe I should write another post about this topic.

You can change Path Selection Policy by several ways and as always the best option depends on your specific requirements and constrains.

However, if you have only one  instance of some storage type connected to your ESX hosts you can simply change default path selection policy for this particular SATP type. Let's assume you have some LSI storage.

Below is simple esxcli command how to do it ...

esxcli storage nmp satp set --default-psp=VMW_PSP_MRU --satp=VMW_SATP_LSI

... and then your default PSP for VMW_SATP_LSI is now VMW_PSP_MRU

One thing you must be aware … if you in the past explicitly changed any devices (disks) to another path selection policy then default PSP will not change even you have another default path policy. There is not esxcli mechanism how to change devices back to accept  default PSP for particular SATP type. Only solution is to edit /etc/vmware/esx.conf

All previous explicit changes are written in /etc/vmware/esx.conf  so it is pretty simple to find it and remove these lines form config file.  I silently assume you do such operations in maintenance mode so after ESX reboot all your paths for your devices will follow default SATP path selection policy. 

BTW: That’s why I generally don’t recommend to change PSP for particular device when not necessary. Sometimes it is necessary for example for RDM’s participating in MSCS cluster. But usually it is abused by admins and implementation engineers.  I strongly believe it is always better to set default PSP to behave as required.

Do you want to test Heavy Load? Try Heavy Load tool.

Bring your PC to its limits with the freeware stress test tool HeavyLoad. HeavyLoad puts your workstation or server PC under a heavy load and lets you test whether they will still run reliably.

Look at http://www.jam-software.com/heavyload/

Simple UNIX Shell Script for generating disk IO trafic

Here is pretty easy unix shell script for disk I/O generation.

#!/bin/sh
dd_threads="0 1 2 3 4 5 6 7 8 9"
finish () {
  killall dd
  for i in $dd_threads
  do
    rm /var/tmp/dd.$i.test
  done
  exit 0;
}
trap 'finish' INT
while true
do
  for i in $dd_threads
  do
    dd if=/dev/random of=/var/tmp/dd.$i.test bs=512 count=100000 &
  done
done

Generated IOs (aka TPS - transaction per second) can be watched by following command

iostat -d -c 100000

Script can be terminated by pressing CTRL-C.

Calculating optimal segment size and stripe size for storage LUN backing vSphere VMFS Datastore

Colleague of mine (BTW very good Storage Expert) asked me what is the best segment size for storage LUN used for VMware vSphere Datastore (VMFS). Recommendations can vary among storage vendors and models but I think the basic principles are same for any storage.

I found IBM RedBook [SOURCE: IBM RedBook redp-4609-01] explanation the most descriptive, so here it is.

The term segment size refers to the amount of data that is written to one disk drive in anarray before writing to the next disk drive in the array, for example, in a RAID5, 4+1 array with a segment size of 128 KB, the first 128 KB of the LUN storage capacity is written to the first disk drive and the next 128 KB to the second disk drive. For a RAID1, 2+2 array, 128 KB of an I/O is written to each of the two data disk drives and to the mirrors. If the I/O size is larger than the number of disk drives times 128 KB, this pattern repeats until the entire I/O is completed. For very large I/O requests, the optimal segment size for a RAID array is one that distributes a single host I/O across all data disk drives. 

The formula for optimal segment size is:
LUN segment size = LUN stripe width ÷ number of data disk drives 

For RAID 5, the number of data disk drives is equal to the number of disk drives in the array minus 1, for example:
RAID5, 4+1 with a 64 KB segment size = (5-1) * 64KB = 256 KB stripe width 

For RAID 1, the number of data disk drives is equal to the number of disk drives divided by 2, for example:
RAID 10, 2+2 with a 64 KB segment size = (2) * 64 KB = 128 KB stripe width 

For small I/O requests, the segment size must be large enough to minimize the number ofsegments (disk drives in the LUN) that must be accessed to satisfy the I/O request, that is, to minimize segment boundary crossings. 

For IOPS environments, set the segment size to 256KB or larger, so that the stripe width is at least as large as the median I/O size. 

IBM Best practice: For most implementations set the segment size of VMware data partitions to 256KB.

Note: If I decrypting IBM terminology correctly then IBM mentioned term "stripe width" is actually "data stripe size". We need to clear terminology because normally is the term "stripe width" used as number of disks in RAID group. "Data stripe size" is payload without the parity. The parity is stored on another segment(s) dependent on selected RAID level.

For clear understanding terminology I've created  RAID 5 (4+1) segment/stripe visualization depicted bellow.

RAID 5 (4+1) striping example

Even I found this IBM description very informative I'm not sure why they recommend to use segment size 256KB for VMware. It is true that the biggest IO size issued from ESX can be by default 32MB because bigger IOs issued from guest OS ESX splits into more IOs (for more information about big IO split see this blog post). However the most important is IO size issued from guest OSes. If you want to monitor max/average/median IO size from ESX you can use tool vscsiStats already included in ESXi for such purpose. It allows you to show histogram which is really cool (for more information about vscsiStats read this excellent blog post). So based on all these assumptions and also my own IO size monitoring in the field it seems to me that average IO size issued from ESX is usually somewhere between 32 and 64KB. So let's use 64KB as average data stripe (IO size issued from OS). Then for RAID 5 (4+1) data stripe will be composed from 4 segments and optimal segment size in this particular case should be 16KB (64/4).

Am I right or I missed something? Any comments are welcome and highly appreciated.

Update 2014/01/31:
We are discussing this topic very frequently with my colleague who work as DELL storage specialist. The theory is nice but only the real test can prove any theory. Recently he performed set of IOmeter tests against DELL PV MD3600f which is actually the same array as IBM DS3500. He found that optimal performance (# of IOPS versus response times) is when segment size is as close as possible to IO size issued from operating system. So key takeaway from this exercise is that optimal segment size for example above is not 16KB but 64KB. Now I understand IBM general recommendation (best practice) to use 256KB segment size for VMware workloads as this is the biggest segment size which can be chosen.

Update 2014/07/23:
After more thinking about this topic I've realized that idea to use the segment size bigger than your biggest IO size can make sense from several reasons

• each IO will get single spindle (disk) to handle this IO which will use queues down the route and will be served in spindle latency time which is the minimal one for this single IO, right?
• typical virtual infrastructure environment is running several VMs generating several IOs based on queues available in the guest OS, ESX layer disk scheduler settings (see more here on Duncan Epping blog) so at the end of the day you are able to generate lot of IOPSes by different threads and load is evenly distributed across RAID group

However, please note, that all this discussion was related to legacy (traditional) storage architectures. Some modern (virtualized) storages are doing some magic on their controllers like I/O Coalescing. I/O Coalescing is IO optimization leveraging reordering smaller IO writes to another bigger IO in controller cache and sending this bigger IO down to the disks. This can significantly change segment size recommendations so please try to understand particular storage architecture or follow storage vendor best practices and try to understand the reason of these recommendations in your particular use case. I remember EMC Clariions used IO coalescing into 64KB IO blocks. 

Related resources:

• Aurimas Mikalauskas : Aligning IO on a hard disk RAID – the Theory
• David Pasek : Disk queue depth in an ESXi environment

IOBlazer

IOBlazer is a multi-platform storage stack micro-benchmark. IOBlazer runs on Linux, Windows and OSX and it is capable of generating a highly customizable workload. Parameters like IO size and pattern, burstiness (number of outstanding IOs), burst interarrival time, read vs. write mix, buffered vs. direct IO, etc., can be configured independently. IOBlazer is also capable of playing back VSCSI traces captured using vscsiStats. The performance metrics reported are throughput (in terms of both IOPS and bytes/s) and IO latency.

IOBlazer evolved from a minimalist MS SQL Server emulator which focused solely on the IO component of said workload. The original tool had limited capabilities as it was able to generate a very specific workload based on the MS SQL Server IO model (Asynchronous, Un-buffered, Gather/Scatter). IOBlazer has now a far more generic IO model, but two limitations still remain:

1. The alignment of memory accesses on 4 KB boundaries (i.e., a memory page)
2. The alignment of disk accesses on 512 B boundaries (i.e., a disk sector).

Both limitations are required by the gather/scatter and un-buffered IO models.

A very useful new feature is the capability to playback VSCSI traces captured on VMware ESX through the vscsiStats utility. This allows IOBlazer to generate a synthetic workload absolutely identical to the disk activity of a Virtual Machine, ensuring 100% experiment repeatability.

TBD - TEST & WRITE REVIEW