best practices vLan

Similar Questions

• Dell MD3620i connect to vmware - best practices

  Hello community,

  I bought a Dell MD3620i with 2 x ports Ethernet 10Gbase-T on each controller (2 x controllers).
  My vmware environment consists of 2 x ESXi hosts (each with 2ports x 1Gbase-T) and a HP Lefthand (also 1Gbase-T) storage. The switches I have are the Cisco3750 who have only 1Gbase-T Ethernet.
  I'll replace this HP storage with DELL storage.
  As I have never worked with stores of DELL, I need your help in answering my questions:

  1. What is the best practices to connect to vmware at the Dell MD3620i hosts?
  2. What is the process to create a LUN?
  3. can I create more LUNS on a single disk group? or is the best practice to create a LUN on a group?
  4. how to configure iSCSI 10GBase-T working on the 1 Gbit/s switch ports?
  5 is the best practice to connect the Dell MD3620i directly to vmware without switch hosts?
  6. the old iscsi on HP storage is in another network, I can do vmotion to move all the VMS in an iSCSI network to another, and then change the IP addresses iSCSI on vmware virtual machines uninterrupted hosts?
  7. can I combine the two iSCSI ports to an interface of 2 Gbps to conenct to the switch? I use two switches, so I want to connect each controller to each switch limit their interfaces to 2 Gbps. My Question is, would be controller switched to another controller if the Ethernet link is located on the switch? (in which case a single reboot switch)

  Tahnks in advanse!

  Basics of TCP/IP: a computer cannot connect to 2 different networks (isolated) (e.g. 2 directly attached the cables between the server and an iSCSI port SAN) who share the same subnet.

  The corruption of data is very likely if you share the same vlan for iSCSI, however, performance and overall reliability would be affected.

  With a MD3620i, here are some configuration scenarios using the factory default subnets (and for DAS configurations I have added 4 additional subnets):

  Single switch (not recommended because the switch becomes your single point of failure):

  Controller 0:

  iSCSI port 0: 192.168.130.101

  iSCSI port 1: 192.168.131.101

  iSCSI port 2: 192.168.132.101

  iSCSI port 4: 192.168.133.101

  Controller 1:

  iSCSI port 0: 192.168.130.102

  iSCSI port 1: 192.168.131.102

  iSCSI port 2: 192.168.132.102

  iSCSI port 4: 192.168.133.102

  Server 1:

  iSCSI NIC 0: 192.168.130.110

  iSCSI NIC 1: 192.168.131.110

  iSCSI NIC 2: 192.168.132.110

  iSCSI NIC 3: 192.168.133.110

  Server 2:

  All ports plug 1 switch (obviously).

  If you only want to use the 2 NICs for iSCSI, have new server 1 Server subnet 130 and 131 and the use of the server 2 132 and 133, 3 then uses 130 and 131. This distributes the load of the e/s between the ports of iSCSI on the SAN.

  Two switches (a VLAN for all iSCSI ports on this switch if):

  NOTE: Do NOT link switches together. This avoids problems that occur on a switch does not affect the other switch.

  Controller 0:

  iSCSI port 0: 192.168.130.101-> for switch 1

  iSCSI port 1: 192.168.131.101-> to switch 2

  iSCSI port 2: 192.168.132.101-> for switch 1

  iSCSI port 4: 192.168.133.101-> to switch 2

  Controller 1:

  iSCSI port 0: 192.168.130.102-> for switch 1

  iSCSI port 1: 192.168.131.102-> to switch 2

  iSCSI port 2: 192.168.132.102-> for switch 1

  iSCSI port 4: 192.168.133.102-> to switch 2

  Server 1:

  iSCSI NIC 0: 192.168.130.110-> for switch 1

  iSCSI NIC 1: 192.168.131.110-> to switch 2

  iSCSI NIC 2: 192.168.132.110-> for switch 1

  iSCSI NIC 3: 192.168.133.110-> to switch 2

  Server 2:

  Same note on the use of only 2 cards per server for iSCSI. In this configuration each server will always use two switches so that a failure of the switch should not take down your server iSCSI connectivity.

  Quad switches (or 2 VLAN on each of the 2 switches above):

  iSCSI port 0: 192.168.130.101-> for switch 1

  iSCSI port 1: 192.168.131.101-> to switch 2

  iSCSI port 2: 192.168.132.101-> switch 3

  iSCSI port 4: 192.168.133.101-> at 4 switch

  Controller 1:

  iSCSI port 0: 192.168.130.102-> for switch 1

  iSCSI port 1: 192.168.131.102-> to switch 2

  iSCSI port 2: 192.168.132.102-> switch 3

  iSCSI port 4: 192.168.133.102-> at 4 switch

  Server 1:

  iSCSI NIC 0: 192.168.130.110-> for switch 1

  iSCSI NIC 1: 192.168.131.110-> to switch 2

  iSCSI NIC 2: 192.168.132.110-> switch 3

  iSCSI NIC 3: 192.168.133.110-> at 4 switch

  Server 2:

  In this case using 2 NICs per server is the first server uses the first 2 switches and the second server uses the second series of switches.

  Join directly:

  iSCSI port 0: 192.168.130.101-> server iSCSI NIC 1 (on an example of 192.168.130.110 IP)

  iSCSI port 1: 192.168.131.101-> server iSCSI NIC 2 (on an example of 192.168.131.110 IP)

  iSCSI port 2: 192.168.132.101-> server iSCSI NIC 3 (on an example of 192.168.132.110 IP)

  iSCSI port 4: 192.168.133.101-> server iSCSI NIC 4 (on an example of 192.168.133.110 IP)

  Controller 1:

  iSCSI port 0: 192.168.134.102-> server iSCSI NIC 5 (on an example of 192.168.134.110 IP)

  iSCSI port 1: 192.168.135.102-> server iSCSI NIC 6 (on an example of 192.168.135.110 IP)

  iSCSI port 2: 192.168.136.102-> server iSCSI NIC 7 (on an example of 192.168.136.110 IP)

  iSCSI port 4: 192.168.137.102-> server iSCSI NIC 8 (on an example of 192.168.137.110 IP)

  I left just 4 subnets controller 1 on the '102' IPs for more easy changing future.

• Connecting two 6224 separate batteries best practices LAG?

  Hello

  I wonder how I should configure LAG between two powerconnect 6224 batteries (2 x powerconnect 6224 by battery) for iSCSI against 4 members EQL traffic, I intend to use the 4 ports of each stack (stack cross-possible LAG on 6224?) and equallogic documentation leaves me in two minds when it wants to run LACP or not?

  I wonder what reviewed best practices in this scenario?

  Cross-stack, LACP and no PLEASE or am I better of without LACP?

  Thanks in advance

  Cree

  The ports connecting the two piles together will be configured differently than the ports of connection of the battery to the EQL appliance. During the connection of the control unit switch EQL is when you might want to disable STP on that specific port.

  «Do not use of Spanning Tree (STP) on switch ports that connect to the terminal nodes (iSCSI initiators or storage array network interfaces).» However, if you want to use STP or Rapid STP (preferable to STP), you must enable port settings available on some switches that allow the port immediately transition to PLEASE State reference to link up. This feature can reduce network interruptions that occur when devices to restart, and should only be enabled on switch ports that connect the nodes. "

  With the network cards on the EQL devices according to me, there is only one active port, and the other is pending. So on the switch ports that are plug on the EQL will be in access mode for your iSCSI VLANS. Maybe someone more about EQL can chime to confirm.

  Here are some good white pages.

  www.dell.com/.../Dell_EqualLogic_%20iSCSI_Optimization_for_Dell_Power_onnect_%20Switches.pdf

  docs.danielkassner.com/.../ISCSI_optimization_EQL.pdf

  www.Dell.com/.../EQL-8024f-4-Switch.pdf

• Best practices with streams WCCP of WAAS

  Hello

  I have a module WAAS SRE 910 in 2911 router that intercepts packets this router with WCCP.

  All packets are received by the external interface (gi 0/2, connected to a switch with port configured in vlan WCCP) and are sent back to the router via the internal interface (IG 1/0 connected directly to the router):

  WAAS # sh interface IG 1/0

  Internet address: 10.0.1.1

  Subnet mask: 255.255.255.0

  State of the admin: to the top

  Operating status: running

  Maximum transfer unit size: 1500

  Entry errors: 0

  Entry packets dropped: 0

  Packets received: 20631

  Output errors: 0

  Output packets dropped: 0

  Load interval: 30

  Input rate: 239 bps, 0 packets/s

  Output: 3270892 bps, 592 packets/s

  Packets sent: 110062

  Auto-negotiation: on

  Full Duplex: Yes

  Speed: 1000 Mbit/s

  WAAS # sh interface 2/0 gi

  Internet address: 10.0.2.1

  Subnet mask: 255.255.255.0

  State of the admin: to the top

  Operating status: running

  Maximum transfer unit size: 1500

  Entry errors: 0

  Entry packets dropped: 0

  Packets received: 86558

  Output errors: 0

  Output packets dropped: 0

  Load interval: 30

  Input rate: 2519130 bps, 579 packets/s

  Output rate: 3431 bps, 2 packets/sec

  Packets sent: 1580

  Auto-negotiation: on

  Full Duplex: Yes

  Speed: 100 Mbps

  The default route configured in module WAAS is 0.0.0.0/0 to 10.0.1.254 (interface of the router).

  It would be better that the packets leave module WAAS of the external interface (instead of the internal interface)?

  Is there a best practice recommended by Cisco on this?

  Thank you.

  Stéphane

  Hi Stephane,

  That's right, internal interface means SM1/0.

  The best way is to have traffic at the start of the internal interface module so that we are sure, it hits the redirect exclude statement and that we do not have the crest of the loops because WCCP.

  Kind regards

  Nicolas

• Best practices for network configuration of vSphere with two subnets?

  Well, then I'll set up 3 ESXi hosts connected to storage shared with two different subnets. I configured the iSCSI initiator and the iSCSI with his own default gateway - 192.168.1.1 - targets through a Cisco router and did the same with the hosts configured with its own default gateway - 192.168.2.2. I don't know if I should have a router in the middle to route traffic between two subnets since I use iSCSI ports linking and grouping of NETWORK cards. If I shouldn't use a physical router, how do I route the traffic between different subnets and use iSCSI ports binding at the same time. What are the best practices for the implementation of a network with two subnets vSphere (ESX host network: iSCSI network)? Thank you in advance.

  Install the most common iSCSI would be traffic between hosts and

  the storage is not being routed, because a router it could reduce performance.

  If you have VLAN 10(192.168.1.0/24) iSCSI, VLAN 20 (192.168.2.0/24) ESX

  MGMT and VLAN 30 (192.168.3.0/24) comments VMs and VLAN 40 (192.168.4.0/24)

  vMotion a deployment scenario might be something like:

  NIC1 - vSwitch 0 - active VMK (192.168.1.10) MGMT, vMotion VMK (192.168.4.10)

  standby

  NIC2 - vSwitch 1 - current (VLAN30) guest virtual machine port group

  NIC3 - vSwitch 2 - active VMK1 (192.168.1.10) iSCSI

  NIC4 - vSwitch 2 - active VMK2 (192.168.1.11) iSCSI

  NIC5 - vSwitch 1 - current (VLAN30) guest virtual machine port group

  NIC6 - vSwitch 0 - MGMT VMK (192.168.2.10) standby, vMotion

  VMK (192.168.4.10) active

  You would place you on VLAN 10 storage with an IP address of something like target

  192.168.1.8 and iSCSI traffic would remain on this VLAN. The default value

  gateway configured in ESXi would be the router the VLAN 20 with an ip address of

  something like 192.168.2.1. I hope that scenario help set some options.

  Tuesday, June 24, 2014 19:16, vctl [email protected]>

• Design by using NetApp's best practices

  I am preparing for my VCP5 and I read the new book by Scott Lowe. the book describes how the traffic should be isolated. your vMotion, vmkernal, etc., but in many organizations, I see the NetApp with some of data warehouses and a few LUNS to CIFS share LUNS. I guess you can have your vmMotion on a VLAN separated, but would not safer just configure a windows VM file server to host your files? In freenas and openfiler forums, they stress is not to run their software in virtual machines in a production environment.   Physical separation would be better then just a VLAN? I was inking and correct me if I'm wrong. I think the CIFS shares in a virtual hosting machine would SAN, vMotion, vmkernal, most reliable if you have redundant switches on both sides VMware hosts. So if your kernel switches drop your vmware environment will not drop.

  > traffic must be isolated.

  Yes, the network traffic must be split on networks separated for various reasons, including performance and safety.

  > NetApps with MON a few for data warehouses and a few LUNS to CIFS share.

  Yes, if you have a NetApp file server you can block-level storage server as FCP or iSCSI, CIFS or NFS file-level storage.

  > I guess you can have your vmMotion on a VLAN separated, but would not safer just configure a windows VM file server to host your files?

  OK, you lost me.  Yes, you must separate the vMotion traffic to enhance the performance and because the vMotion traffic is not encrypted.

  I don't see where you're going for vMotion to a Windows file server?

  However, if you are referring to, why don't you your NetApp instead of Windows CIFS Server:

  You don't need to patch and reboot the NetApp at least once a month.

  Performance is better

  You don't need to buy a Windows license and then maintain Windows

  Snapshots.  NetApp has the best shots in the business.  When your Windows I/O high, or just typing box because it of Tuesday and removes all of your VSS snapshots you really wish you had a NetApp.

  > In the forums of freenas and openfiler, they stress is not to run their software in virtual machines in a production environment.

  Note that there are a ton of storage there equipment running as VMs and server NFS for shared storage, including left and they have been stable for years.

  > Physical separation would be better then just a VLAN?

  Yes, if you have the infrastructure.  When it comes to the first time I've seen reference you VLAN?  Are you talking about now the NetApp as the series 2020 with two network cards where you need to carry all traffic (managent, CIFS and iSCSI) through them via VLAN?

  Like this: http://sostechblog.com/2012/01/08/netapp-fas2xxx-fas3xxx-2-nic-ethernet-scheme/

  > I was inking and correct me if I'm wrong. I think the CIFS shares in a virtual hosting machine would SAN, vMotion, vmkernal, most reliable

  CIFS is nothting to do with SAN, vMotion or VMkernel.  CIFS (SMB) is the protocol used mainly by Windows file sharing

  > If you have redundant switches on both sides of the VMware hosts. So if your kernel switches drop your vmware environment will not drop.

  You always want to redundant switches.  No single point of failure is the best practice.

• Best practices Networking ESXi 5

  Good afternoon.

  Hate me help, docks as best practices para uma infraestrutura of rede receber o Vmware (ESXi 5 + vCenter vConverter) autonomous.

  Switch, VLAN, NIC?

  Obrigado

  Olá amigo, good afternoon.

  Leave aqui um trecho do livro "Virtualização - Central Datacenter do Componente" em than comenta, sober os conceitos uma rede virtual cloud:

  An e VMware a Savvis sugerem some measures Segurança interesting than sao Speaker aqui operacionais:

  Lado provedor:

  • Manter as redes for sao to leave da infraestrutura virtualizada isoladas. Insulation pode ser feito com os rocking or utilizando segmentacao through VLAN. Outro Método PODES use uma want Kanada Doi methods e o conceito of virtual switch.
  • Manter as redes of members isoladas. Manter as redes of members e isoladas interfaces controlled devidamente.
  • Manter as redes used para verificaram maquinas e tolls IP isoladas em redes roteaveis nao. Estas redes precisam ser rapidas e ao mesmo tempo sao suscetiveis an attacks.
  • Manter as redes back clients isoladas. Estas redes should ser isoladas das redes e members should divertir firewall between as redes para avoid security problems.
  • Fornecer seguro aos recursos nuvem client access. Client OS usually precisam ter access has recursos dentro da e nuvem para isto e Chipre o provedor disponibilize um Portal com encriptacao para Segurança effect.
  • Backups to maintain E restores seguros e consistent. Provedor deve to maintain this appearance basico realize e recursos back to maintain backup um processo rapido restore.
  • Autenticacao strong mechanisms, auditoria e autorização. OS provedores should please has autenticacao segura, access so aos recursos waiting rooms recursos providenciar e.
  • Use models seguros e configuracao gold images do sistema operational e das management. Estas reduzem os problemas com configuracoes inadequadas measures. Also o provisionamento maquinas tolls deve obedecer an entrar em being Produção antes criteria preestabelecidos.
  • Members of recursos para avoid attacks do tipo back.

  Client side:

  • Practices of mercado Seguir para a Segurança sistemas operacionais back. OS administradores should seguir as mesmas rules used no ambiente interno relativas to back sistemas operacionais, como manter os last Segurança no patches of Segurança configuracao gold images.
  • Following to encrypt data. Dados follows sistemas important should mantidos encriptados forse ainda but os possíveis ataque para ser to rede.

  Espero ter helped.

• Best practices of ESXi.

  I'm trying to find out if I implement best practices correctly on ESXi 4.1 U1. Our facility has two on-board NICs and a quad card NETWORK to use for data and management networks. We have fiber for the side storage. If I understand the latest practices should I trunk 6-port then use port groups to manage traffic VLAN where going.

  My question is this does not put excessive pressure on the network card that is used for the traffic of vmotion/vmkernel? If I put the switchport in one VLAN specific datagrams wouldn't get inspected and abandoned by the switch asics as opposed to the ESXi host to have to deal with them? I don't know if traffic would have a significant impact, so be it, but wanted to check before the implementation...

  Thank you
  Froggy

  I guess that you are concerned about air on the ESXi host performance, because it must deal with VLAN tags.

  Don't worry about this. The overhead is minimal and modern NICS also have material unloading of functions for VLAN that will be used by ESX in many cases of handling.

  However, remember that you have no safe the uplinks on the physical switch! Just create a virtual switch and add all the rising physical their. ESXi them then use all function of the active/standby/unused attribute that you can set by group of virtual ports.

  Andreas

  - Blog of experience front of VMware

• Separate management / VMotion Best Practice?

  We're heading to 4.0 ESX ESXi 4.1.  Our servers have 4 physical Gigabit NIC.

  On ESX 4.0, we lack 2 vSwitches:

  vSwitch0

  Service Console - Active vmnic0 - vmnic3 watch

  VMkernel - Active vmnic3 - vmnic0 eve

  (Unique network interface cards / IPs by function)

  vSwitch1

  Port VM - vmnic1 and vmnic2 active groups

  (Several VLANS to resources shared)

  With the changes in ESXi, is recommended to separate the management of VMotion as we did with ESX?  Notice that we use the same subnet for these two functions.

  Personally, I would prefer combining Management and VMotion.  VMotion will not only benefit an additional NIC usage, especially with the multiple simultaneous VMotions?  At the same time, it seems not that management traffic would be impeded to the point of needing to separation, as we use the same subnet.  In addition, security should not be a problem, since the new, we use the same subnet for management and VMotion.

  Your configuration is consistent with "best practices". I prefer separate management taffic VMkernel myself, even if it will cost me some performances of vMotion.

  ---

  MCITP: SA + WILL, VMware vExpert, VCP 3/4

  http://blog.vadmin.ru

• The best practices with only 4 cards network network

  I'm used to esx with fiber channel without configuration.  This will be my first experience connecting esx to an iscsi san.

  I just have 4 network cards on the server and wonder what is the best practice.

  I think that 2 of the 4 network cards will be dedicated to the San for HA.

  But the other 2 types of throw me a loop for.

  I know I want to console, vmotion and trunk for VLANs, but I don't think I can have all 3 on the same vswitch?

  Am I stuck with having the console and vmotion on a network card and the trunk of the other?  IE - no failover?

  Is there anyway to get console, vmotion and trunk at all exist together?

  Thank you in advance.

  If it were an election, I vote for Josh26

  Just a few additions to his recommendation:

  vSwitch0:

  2 natachasery connected to the ports of the trunk on the pSwitch (vmnic0, vmnic1)

  Service - VLANxxx - Portgroup (vmnic0, vmnic1 active standby) Console

  VMkernel Portgroup for vMotion - VLANyyy - (vmnic0 vmnic1 standby, active)

  Machine Virtual Network - VLANzzz - (vmnic0, vmnic1 active standby)

  vSwitch1:

  2 natachasery (vmnic2, vmnic3)

  VMkKernel Portgroup for iSCSI

  With the above configuration you separate iSCSI traffic, what is most important.

  SC and the VM network share a teddy and vMotion has its dedicated Teddy.

  Everything is redundant.

  André

• NIC Teaming best practices

  Hello

  I have 1 server that has ports gigabit 8 inside. There will be 6 VLAN (including VLAN ID 4095) inside this ESX Server. Is it best practice to all gigabit 8 ports combine and link to vSwitch0 (default) and create the Group of ports by VLAN?

  Kind regards

  T.S.

  is absolutely not a stupid question

  Only for safety and performance. You can use configuration grouping in both cases. (2 NICs for sc and vmotion and 4 NICs for vlan tag vm)

  Keep this is to undertatnd that it is not an absolutely answare, but only a best practice, and that's OK as your design.

  I thank Alberto

• looking for a networking best practices publication

  The following quote re ESX host network configuration came from one of our technical pre-sales representatives IBM in the Vancouver area...

  "IBM and VMware have long have jointly published a best practice of the 2 Ethernet interfaces (1 Gb/s each) have teamed up for a redundancy and bandwidth - separated into 3 VLANs - to serve the three channels (management, vmotion and vmnetwork).

  Someone at - it a link to this document?  I can't seem to find it, and I really want to read.

  Thank you.

  Hello

  Did not see the document you are looking for, but if you want to learn more about some different network configurations have a look here:

  http://www.NetworkWorld.com/community/blog/7119

  -Pal-André

• Implementation of best practices

  We have a budget tight but would like to be considered for implementation, as many best practices of VMware as possible. We realize that, with our budget, we will not be able to implement many of them but I would like to get the most, most critics made. What do you consider the ABSOLUTE MUST have best practices to an environment that includes the following:

  6 hosts ESX3.5 and 1 VC2.5 with the license server Center. Each host has 6 physical network interface cards. We have licenses for vMotion, DRS, and HA. Our current setup is:

  vSwitch0 is home to a group of virtual computer and Console of Service ports with 2 network cards

  vSwitch1 is home port for VM group

  vSwitch3 is home to port VM group

  vSwitch4 is a dedicated vMotion network

  Physically, our hosts are connected to a single physical Nortel SwitchStack affected ports to the VLAN

  We have our virtual machines on different subnets 3

  This infrastructure is inside our firewall. We have a demilitarized zone, but it is not virtual.

  Looking for candid suggestions.

  The vSwitch is an object that exists in the memory of an ESX server.  Therefore, the exchanges on this vSwitch belong within this memory object.  This vSwitch keeps his stats, the mac entity table.  With the help of a 2nd vSwitch creates a 2nd object that does not interact/interface with the first.  Then, it adds another layer of separation between the information.  Gaps are a good thing for this reason.

  -KjB

• Best practices Apple ID

  I help the family members and others with their Apple products. Probably the problem number one revolves around Apple ID I saw users follow these steps:

  (1) share IDs among the members of the family, but then wonder why messages/contacts/calendar entries etc are all shared.

  (2) have several Apple IDs willy-nilly associated with seemingly random devices. The Apple ID is not used for anything.

  (3) forget passwords. They always forget passwords.

  (4) is that I don't really understand. They use an e-mail from another system (gmail.com, hotmail.com, etc) as their Apple ID. Invariably, they will use a different password for their Apple ID than the one they used for other email, so that they are constantly confused about which account to connect to.

  I have looked around for an article on best practices for creating and using Apple ID, but could not find such a position. So I thought I would throw a few suggestions. If anyone knows of a list or wants to suggest changes/additions please feel free. Here are the best practices for normal circumstances, i.e. not cooperate accounts etc.

  1. every person has exactly 1 Apple ID.

  2. do not share Apple ID - share content.

  3. do not use an email address of another counts as your Apple ID.

  4. When you create a new Apple ID, don't forget to complete the secondary information to https://appleid.apple.com/account/manage. It is EXTREMELY important questions your email of relief and security.

  5. the last step is to collect the information that you entered in a document and save to your computer AND print and store it somewhere safe.

  Suggestions?

  I agree with no. 3, it is no problem with using a addressed no iCloud as the primary ID, indeed, depending on where you set up your ID, you may have no choice but to.

• Best practices Upgrade Path - Server 3 to 5?

  Hello

  I am trying a migration and upgrade of a server in the Profile Manager. I currently run an older mac mini Server 10.9.5 and Server 3 with a vast installation of Profile Manager. I recently successfully migrated the server itself out of the old mac mini on a Xserve end 2009 of cloning the drive. Still of double controls everything, but it seems that the transition between the mini and the Xserve was successful and everything works as it should (just with improved performance).

  My main question is now that I want to get this software-wise at day and pass to the Server 5 and 10.11. I see a lot of documentation (still officially Apple) best practices for the upgrade of the Server 3 to 4 and Yosemite, but can't find much on the Server 5 and El captain, a fortiori from 3 to 5. I understand that I'll probably have to buy.app even once and that's fine... but should I be this staging with 10.9 to 10.10 and Server 4... Make sure that all is well... and the jump off 10.11 and Server 5... Or is it 'safe' (or ok) to jump 3 to 5 Server (and 10.9.5 to 10.11.x)? Obviously, the AppStore is pleased to make the jump from 10.9 to 10.11, but once again, looking for best practices here.

  I will of course ensure that all backups are up-to-date and make another clone just before any which way that take... but I was wondering if someone has made the leap from 3-5... and had things (like the Profile Manager) still work correctly on the other side?

  Thanks for any info and/or management.

  In your post I keep the Mini running Server 3, El Capitan and Server 5 install the Xserve and walk through setting up Server 5 by hand. Things that need to be 'migrated' as Open directory must be handled by exporting the mini and reimport on Xserve.

  According to my experience, OS X Server facilities that were "migrated" always seem to end up with esoteric problems that are difficult to correct, and it's easier to adopt the procedure above that to lose one day try.

  YMMV

  C.