Windows Server Analysis - as of ZVPS 4.3
This page explains managing performance for Microsoft Windows servers. The technology used by zVPS for distributed servers is standard SNMP. Microsoft servers have implemented both the standard network MIB in SNMP, as well has the HOST MIB. The HOST MIB has also been implemented by all Linux SNMP variants as well, and is extremely useful for general management of distributed servers.
Management Definition
The data provided for windows servers meets the Velocity Software requirements for performance management, meaning that all 4 parts of performance management requirements are met:
- Performance Analysis means that metris are available to understand the workloads resource requirements.
- Capacity Planning means the metrics are maintained for historical analysis in the zVPS performance database.
- Operations Support allows any metric to have an alert maintained as part of the standard zVPS operations.
- Chargeback Support provides the ability to monitor servers to the process level to understand how much resource is being consumed.
Host Mib
There are several types of data returned by the Host MIB.
- File System Data: Seeing all the mounted file systems
that the windows servers have, how large they are, how full they are
means that alerts can be set for disk "almost full" conditions to ensure
applications do not fail for out of space conditions. Alerts can be
set to check for file systems being mounted as well.
Screen: ESAHST2 Velocity Software - VSIVM4 ESAMON 4.303 04/19 20:45-20:46 1 of 3 LINUX HOST Storage Analysis Report CLASS * NODE WIN* 2828 0414C7 <-Utilization-> Node/
Pct Alloc Storage Time Group Size Used Full Units R/W Description -------- -------- ----- ----- ---- ----- ---- -------------------------------- 20:46:00 win64 8191 2601 31.7 66K Physical Memory 16381 3124 19.1 66K Virtual Memory 954K 694K 72.8 4K r/w L:\ Label:Linux64 Serial Number 12292 8062 65.6 4K r/w D:\ Label:FACTORY_IMAGE Serial 941K 302K 32.1 4K r/w C:\ Label:HP Serial Number ad0b win10 8182 1250 15.3 66K Physical Memory 16374 1143 7.0 66K Virtual Memory 476K 149 0.0 4K r/w E:\ Label:Edrive Serial Number 3075 3075 100 2K r/o D:\ Label:GRMCULXFRER_EN_DVD Se 305K 39646 13.0 4K r/w C:\ Label:Cdrive Serial Number - Hardware Configuration: The HOST MIB also provides the
hardware configuration and the status. The list of devices online
seems to be complete, the status seems to be incomplete at this time.
Screen: ESAHST3 Velocity Software - VSIVM4 ESAMON 4.303 04/19 20:46-20:47 1 of 1 LINUX HOST Device Report NODE WIN* 2828 0414C7 <------------------- System Device -------------------------> Server Index Status Errors Description -------- ----- ------- ------ ---------------------------------------- win64 62 unknown 0 Fixed Disk 61 unknown 0 Fixed Disk 21 unknown 0 USB Wireless 802.11 b/g Adaptor 10 unknown 0 Software Loopback Interface 1 9 running 0 Unknown Processor Type 8 running 0 Unknown Processor Type 7 running 0 Unknown Processor Type 6 running 0 Unknown Processor Type 5 running 0 EPSON PC-FAX driver 4 running 0 Microsoft Shared Fax Driver 3 running 0 Microsoft XPS Document Writer 2 running 0 EPSON WF-7520 Series 1 running 0 Xerox Phaser 6180DN PS win10 10 unknown 0 Microsoft ISATAP Adapter 9 unknown 0 USB Wireless 802.11 b/g Adaptor 8 unknown 0 Software Loopback Interface 1 7 running 0 Unknown Processor Type 6 running 0 Unknown Processor Type 5 running 0 Unknown Processor Type 4 running 0 Unknown Processor Type 3 running 0 Microsoft Shared Fax Driver 2 running 0 Microsoft Print To PDF 1 running 0 Microsoft XPS Document Writer v4
- Process Table: Windows has an "interesting" characteristic
in how it updates the SNMP data. The process table is available, but the
performance metrics are not updated at the zVPS preferred time frame.
With
the standard one minute data collection standard with zVPS, MS Servers
appear inconsistent. The reality is that the SNMP process table data
is updated every 2-3 minutes. If you monitor an MS server every 5
minutes, there would be data consistently. But monitoring
every minute means there are gaps in the data. From this data, one
can see the complete process table, but when looking at the active processes
over time, one must recognize the MS deficiency. zVPS shows active
processes - defined as those using more than .1% of a cpu. So in this example
of one server, we know the active processes, but not in the exact detail
as we would get from linux servers.
There is an "Idle" process maintained by Windows in the process table that gets credit for unused CPU. This process shows up every 2-3 minute and those intervals then represent correctly the previous 2-3 minutes. With the idle process, adding up all the processes adds up to the total system capability. This is a 4-cpu server, and typical of the environment, utilization is very low - but it IS measurable...Screen: ESAHST1 Velocity Software - VSIVM4 ESAMON 4.303 04/19 20:41-20:48 1 of 1 LINUX HOST Software Analysis Report NODE WIN64 2828 0414C7 <--Software Program---------->
CPU Storage(K) Time Node Name ID Type Status Total Intrval Pct Current -------- -------- -------- ----- ------- ------- ----- ------- ----- --------- 20:49:00 win64 *Totals* 0 0 476.66 265 1868M System I 1 OpSys Running 825K 474.05 263 24576 System 4 OpSys Running 1598 0.53 0.29 21762K svchost. 512 Applic Running 3919 0.45 0.25 22020K winlogon 720 Applic Running 46 0.03 0.02 10727K nvSCPAPI 940 Applic Running 34 0.03 0.02 6320K svchost. 968 Applic Running 1053 0.03 0.02 219M spoolsv. 1624 Applic Running 75 0.05 0.03 15593K schedul2 1776 Applic Running 142 0.11 0.06 8966K nsbu.exe 2316 Applic Running 4689 0.50 0.28 22544K snmp.exe 2528 Applic Running 1200 0.87 0.48 8495K 20:48:00 win64 *Totals* 0 0 1.84 1.53 1872M Skype.ex 4720 Applic Running 86 0.14 0.12 133M SkypeBro 4948 Applic Running 5 0.02 0.02 57209K TrueImag 5132 Applic Running 2918 1.68 1.40 24367K 20:46:00 win64 *Totals* 0 0 479.55 400 1871M System I 1 OpSys Running 824K 473.56 395 24576 System 4 OpSys Running 1597 0.43 0.36 21762K svchost. 512 Applic Running 3919 0.37 0.31 21983K csrss.ex 556 Applic Running 451 0.02 0.02 5382K winlogon 720 Applic Running 45 0.02 0.02 10727K svchost. 848 Applic Running 55 0.10 0.08 10580K nvSCPAPI 940 Applic Running 34 0.04 0.03 6320K svchost. 968 Applic Running 1053 1.48 1.23 219M schedul2 1776 Applic Running 142 0.06 0.05 8970K nsbu.exe 2316 Applic Running 4688 0.75 0.63 26735K snmp.exe 2528 Applic Running 1199 0.89 0.74 8483K Skype.ex 4720 Applic Running 86 0.11 0.09 133M SkypeBro 4948 Applic Running 5 0.03 0.02 56361K TrueImag 5132 Applic Running 2917 1.66 1.38 24367K 20:44:00 win64 *Totals* 0 0 478.96 399 1877M System I 1 OpSys Running 824K 473.93 395 24576 System 4 OpSys Running 1597 0.61 0.51 21762K
Network Mib
The network data provided by MS servers is standard, and allows for complete network management with no additional agents.
- Transport Data: The transport layer data shows what the high
level TCP and UDP datas have, how large they are, how full they are. That
means alerts can be set for disk "almost full" conditions to ensure
applications do not fail for out of space conditions. Alerts can be
set to check for file systems being mounted as well.
Screen: ESATCP1 Velocity Software ESAMON 4.301 04/18 09:02-10:06 1 of 2 TCPIP Transport Layer Data CLASS * NODE WIN* 2828 0314C <----- TCP Connections ------>
And the UDP data:Node/ Curr <---Segments Transmitted---> Time Group Conn Activ Pass Fails Resets Input Outpt ReTrn InErr Rsts -------- -------- ---- ----- ----- ----- ------ ----- ----- ----- ----- ---- 10:06:00 win64 24 0.0 0 0 0 0.7 0.6 0 0 0 10:05:00 win64 24 0.0 0 0 0 10.4 5.7 0 0 0 10:04:00 win64 22 0.2 0 0 0 6.2 5.3 0 0 0 10:03:00 win64 22 0.2 0 0 0.0 8.2 7.5 0 0 0 10:02:00 win64 25 0.3 0 0 0 33.9 23.3 0 0 0 Screen: ESATCP1 Velocity Software ESAMON 4.301 04/18 09:04-10:05 2 of 2 TCPIP Transport Layer Data CLASS * NODE W* 2828 0314C <--UDP Datagrams per second--> Node/ Total Total <----Errors---> Time Group Input Output NoAppl Other -------- -------- ----- ------ ------ ----- 10:05:00 win64 2.0 2.1 0.4 1.7 10:04:00 win64 1.8 1.8 0.6 1.7 10:03:00 win64 3.1 3.4 0.6 1.6 10:02:00 win64 5.3 5.4 0.4 1.4
- Internent Protocol Data: The IP layer data shows the high
level IP traffic in and out.
Screen: ESATCP2 Velocity Software ESAMON 4.301 04/18 10:00-10:10 1 of 2 TCPIP Internetwork Layer Data CLASS * NODE WIN* 2828 0314C
Node/ Time Group Total Fwrd Dlvrd Hdr Addr Prot Other Reqst NoRte Other -------- -------- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- 10:09:00 win64 22.02 0 26.98 0 0.13 0 1.92 17.03 0 0 10:08:00 win64 22.10 0 30.08 0 0.22 0 2.01 19.58 0 0 10:07:00 win64 3.64 0 3.39 0 0.12 0 1.65 2.01 0 0 10:06:00 win64 4.73 0 4.30 0 0.22 0 1.59 2.94 0 0 10:05:00 win64 3.65 0 3.13 0 0.22 0 1.22 2.31 0 0 10:04:00 win64 13.44 0 14.10 0 0.13 0 2.11 7.84 0 0 10:03:00 win64 6.54 0 9.68 0 0.17 0 2.32 7.14 0 0 10:02:00 win64 8.64 0 13.05 0 0.20 0 2.21 10.96 0 0 10:01:00 win64 32.25 0 40.65 0 0.13 0 1.77 28.71 0 0 - Interface Data: The hardware layer is more interesting
in that traffice by interface is shown.
Screen: ESATCP4 Velocity Software ESAMON 4.301 04/18 10:13-10:14 1 of 2 TCPIP Hardware Layer / Interfaces CLASS * NODE WIN64 2828 0314C7
<-Average-> <-Subnet packets / Sec-> Node/ <-Per second-> Packet Size <-Unicast-> Time Group Interface Input Output In Out Input Outpt Input Output -------- -------- --------- ------ ------ ----- ----- ----- ----- ----- ------ 10:14:00 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 Intel(R) 2312 2534 153 476 4.79 5.28 10.30 0.05 win64 Intel(R) 2311 2534 153 476 4.79 5.28 10.30 0.05 win64 VirtualBo 0 4.06 . 243 0 0 0 0.02 win64 VirtualBo 0 4.06 . 243 0 0 0 0.02 win64 VirtualBo 0 4.06 . 243 0 0 0 0.02 win64 WAN Minip 0 0 . . 0 0 0 0 win64 WAN Minip 0 0 . . 0 0 0 0 win64 WAN Minip 0 0 . . 0 0 0 0 win64 Intel(R) 2311 2533 153 476 4.79 5.28 10.30 0.05 win64 Microsoft 0 0 . . 0 0 0 0 win64 Intel(R) 2311 2533 153 476 4.79 5.28 10.30 0.05 win64 USB Wirel 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 VirtualBo 0 4.06 . 243 0 0 0 0.02 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 Microsoft 0 0 . . 0 0 0 0 win64 WAN Minip 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0 win64 USB Wirel 0 0 . . 0 0 0 0
Microsoft Review
MS data using the SNMP data source is not as good as the Linux data source. It does provide the ability to know processor utilization, file systme utilization and provides some level of configuration knowledge. This is likely sufficient for most MS environments.
Using the standard snmp interface means that there are no extra agents or software that needs to be installed (and tested). Please send questions or comments to Barton....