E20-547 Brain Dump

2012-10-01T06:51:00.000-07:00

Peer control station connects to eth1

The customer network connects to eth0

Fmp protocol is used to exchange info between the mpfs and the vnx array

DNS service resolution resolves service names instead of computer names. DNS returns a list of machines that run a specific service, such as LDAP and Kerberos

DNS are defined to data mover using the server_DNS command

You must configure more then one DNS server for each data mover when:

· Not in same windows forest

· Not served by the same DNS server

UDP is the preferred protocol for DNS

Protocolwill switch to tcp when message is larger than 512 bytes

2 ntp servers recommended server_date to set ntp services

To migrate between domains use server_cifs –migrate, -replace

The CIFs server created on a physical data mover with no specified interfaces becomes the default server. It is associated with all unused interfaces and subsequent interfaces attached.

Server_export command to export shares

Unicode 3.0 standard

Share name limited to 12 characters with Unicode supports 80 characters

Comment length is 256 bytes /256 ASCII

Best practice to enable Unicode at start cannot go back to ASCII changing to Unicode after installation can cause interruption

Quotas can be configured using the CLI of vnx, unisphere or windows server interface

Activate quotas before populating the filesystem when quotas are first activated the whole filesystem is unavailable while the quota initiation process sees the fs

VNXe 3100/3300 – File Iscsi block

VNX 5100 – FC Block only

VNX5300/5500/5700/7500 – File and Block

VG2/VG8 – file only

Control Station runs red hat 5

Ifconfig eth0 (will show eth0 properties and ip address

Quotas can be managed via CLI or unisphere gui or windows hosts

Anit virus is part of the VNX event enabler infrastructure

SAN Copy not part of any suites comes as standard

VNX Protection suites	Total protection	Total efficiency
fastsuite	N	y
Security and compliance suite	N	Y
Local protection suite	Y	Y
Remote Protection suite	Y	Y
Application protection suite	Y	Y

Fast suite

VNX Fast VP (Less inactive data goes to slower drives more active data goes into cache or faster drives)

VNX Fast Cache – up to 2Tb of cache

Unisphere analyser

Unisphere quality of service manager

Security and Compliance suite

Event Enabler – auditing sets retention perioads for files

File level retention – WORM

VNX Host Encryption – PowerPath Encryption/Volume Level Encryption/RSA Keys

Local Protection Suite

VNX Snap View - Block snaps 8 snap sessions max

VNX Snapsure – File snaps 16 writable snaps max 96 read only snaps 112 altogether

Recoverpoint/SE – Continuous Data Protection

Remote Protection Suite

VNX Mirrorview – Mirrorview/s 1:1 relationship Mirrorview/a 4:1 relationships

VNX Replicator – Local Copy, Remote Copy, IP based, file replicator, failover, failback

Recoverpoint/SE

Continuous Remote Replication CRR FC-IP data compression check sum verfified

Application protection suite (manages snaps, clones)

Replication manager

Data protection advisor for replication

Data protection advisor (collects > correlates > analyses > presents) alerts, advice, trending

RAID Type	Min	Max	Data protection	Usage
0	3	16	None striped	Use when speed is the main concern
1	2	2	Mirrored	Provides data integrity (Ideal for OS’, Database log files)
1/0	2	16	Mirrored stripes	Performance and protection is the main concern (Largest amount of disk wastage)
3	5 or 9	5 or 9	Parity on single disk
5	3	16	Parity across all disks	Low cost protection is the main consideration good random read, very good sequential read performance
6	4	16	DP across all disks	Low cost protection is the main consideration good random read, very good sequential read performance

Magic bytes – Shed stamp, write stamp, time stamp, checksum

Disks are formatted at 520 bytes per sector, 512 bytes holds accessible data. 8 bytes are used by the array for background verify > checks data and corrects at the SP.

A shed stamp is used with RAID 3/5 and performs parity shedding

A write stamp is used to detect when writes complete which is less than a full stripe used with RAID 5.

A time stamp is used with RAID 5/3 and is used to detect when an incomplete stripe is written.

Checksum is used in all RAID types and is used to detect but not correct errors in the sector.

RAID 6 parity shedding

When a disk fails writes get written to the parity disk making a hsed position, once in a shed position there is no protection reads will then come from the shed.

RAID 6 is row parity and diagonal parity

2 possible shed positions

Sniff Verify (When LUN is idle runs at increased rate) – low priority check on entire storage system, very low impact on performance

Background verify –

· high priority check of a LUN or RG

· can be initiated automatically by the system

· once started it runs to completion and cannot be stopped

· starts automatically if during a trespass the OS determines the data could be inconsistent

X-blade standby’s must have identical i/o modules

1 x-blade standby for 3 primary

Alua uses failover mode 4

i/o redirected to other SP over the CMI channel

unisphere tabs (Dashboard > system>storage>hosts>data protection >settings > support)

ssh to control station

NaviCLI to VNX Block

Storage groups are access logix containers for grouping LUNs

RAID groups limited to 16 disks

RAID Groups can be defragmented except for RAID-6

RAID 6 needs an even number of disks

RAID 1/3/6 cannot be extended

EMC recommends 5 disks in RAID 5 and 8 disks in a RAID 6 in storage pools

SP’s have up to 24Gb of RAM at 1330Mhz DDR3 uses Xeon 5600 CPU’s

NFS, PNFS(NFSv4.1)

CIFs, MPFS (Multi path file system)

Object Storage REST, SOAP – Atmos ve

VNX series gateway is file only

AVNX gateway supports up to four back-end arrays

x-blade for file front end

Storage Processor for block front end

Control station used to configure, manage and upgrade the x-blades and manage failover. Each x-blade enclosure has 2 x-blades running VNX OE and you can have up to 8 x-blades.

x-blades can be configured for N + 1 or N+M

DPE (Disk Processor Enclsure)

SPE (Storage Processor Enclosure)

i/o modules for x-blades

4 x 1Gb base T

2 x 1Gb base T + 2 x 1GbE optical

2 x 10GbE optical

2 x 10GbE Twin ax

i/o modules for SP’s

4 x 8Gb FC

4 x 1Gb base T (iSCSI)

2 x 10GbE Optical (iSCSI)

2 x 10GbE Twin ax (iSCSI)

2 x FCoE

VG2 N + 1

VG2 N + 1, N + M

x-blade failover times

minimal – 15/30 seconds

maximal – 1 minute 50 seconds

FTP and NDMP must be restarted manually after failover

Flash EFD, SSD 3.5” 100Gb, 200Gb 3000 i/o’s per second

SAS 3.5” 300Gb, 600Gb (195 drives/Rack) 10k/15k 140 iop’s (10k) 180 iop’s (15k)

SAS 2.5” 300Gb, 600Gb (500 drives/rack) 10k

NL-SAS 3.5” (195 drives/rack) 1Tb, 2Tb, 3Tb 7.2k 90 iop’s (7.2k)

DAE’s config rules

Max number of enclosures per bus 10

Max number of slots per bus 250

Serial attached SCSi back end (SAS)

6Gb/s = 750Mb/s

4 x 750Mb/s = 3Gb/s

i/o ports on SP slot 4 has 4 port FC i/o card connects to x-blade otherwise other 3 ports used for block connectivity

Control Station eth1 port used to connect to a peer control station port. When there are 2 control stations

eth0 connected to primary x-blade config port

eth2 secondary x-blade

eth3 connects to customer/user network for mgmt unisphere is accessed over this interface.

VNX 5100/5300/5500 use DPE’s

VNX 5700/7500 use SPE’s

During install powerpath determines all paths to a device and builds a table deciding how to route i/o. Powerpath reads table and directs i/o to a native or pseudo device.

Power Path Licensing

· Power Path – unlocks all load balancing and path failover

o Symetrix optimise

o Clarion optimise

o Adaptive optimise for non EMC

· PowerPath SE

o Backend (between switch and array) failover only

o Single HBA supported

o No load balancing

o Up to two paths per device

· PowerPath VE

o Same as powerpath license but only supports w2k8, hyper-v and vSphere

Powerpath load balancing policies

· Sym_opt/clar_opt/adaptive (default)

· RoundRobin – i/o distributed down each path in turn

· Least i/o – i/o is sent down the path with the least i/o in the queue

· Least_blocks – i/o is sent down the path with the least blocks in the queue

· Request – failover only

· No direct – disable pf and lb default for symetrix and clarion with no license

· Base failover – fabric failover

Auto probe – tests for dead paths (uses scsi enquiry commands every 30 seconds)

Auto Restore – tests for restored paths (runs every 5 minutes)

Powerpath can work with cluster software

Power path commands

1. Powermt check

2. Powermt check_registration

3. Powermt config

4. Powermt display

5. Powermt display dev=all

6. Powermt display options

7. Powermt display paths

8. Powermt load

9. Powermt remove

10. Powermt restore

11. Powermt save

12. Powermt set mode

13. Powermt set periodic_autorestore

14. Powermt set policy

15. Powermt set write throttle

16. Powermt set write_throttle_queue

17. Powermt version

Powerpath uses RSA to encrypt data

Powerpath migration enable – non disruptive data migrations but needs replication software

Basic maintenance (use procedure generator)

USM Activites (Unisphere Serve manager)

· Registration

· Generate and view system configs

· Hardware

o Installation (DAE, disk, i/o module, sfp)

o Replacement (helps replace failed hardware)

· Software

o System software (helps selects appropriate software to install)

o Powerlink downloads (disk firmware)

o Disk firmware

· Diagnostics

o Verify storage system (health checks produces reports)

USM can manage VNX, Clariion, celera running DART 5.6

Download generator http://powerlink.emc.com/csweb2/dgweb/VNXpg/

F some components have failed

T – storage system or component is in transition state

U – system is unmanaged

X – storage system is unaccessible

? – unsupported

Enclosure LED

Green (15 drive bay) AC or DC applied outputs in range

Blue (25 bay enclosure) AC or DC applied outputs in range

Off all outputs out of range

Enclosure fault LED

Off all outputs out of range

Amber any enclosure fault, including SP faults

Array is up and running

Power LED on SP is green

Fault LED on SP is off

DPE status is blue or green

First 4 drives are green

USM-tools to check firmware

Identify failed hardware with unisphere

· Enter ip in browser window

· Select storage system from dashboard

· Select system > storage hardware or hardware for file

· Check for failed components

Verify storage system

· Initial system check

· Capturing config data

· Generating issues report

Find failed disk using unisphere > click replace disk (launches USM)

SP events log also shows failed disk

DAE-15 to a DPE-25 or a DAE-25 to DPE15 is allowed

DAE6s 3u 15 drive

DAE5s 2u 25 drive

The maximum amount of enclosures is calculated based on the max number of drives

Create pools > advanced

Pool alerts: percent full threshold

Fast cache: enable disable

RG > advanced

Auto delete after last lun deleted

Expand and defrag priority H,m,L

Allow power savings

Thin luns cannot be used in the reserve lun pool, the write intent log or clone private luns

Thick luns cannot be used with write intent log or clone private luns. Can be used in the reserved lun pool.

Thin and thick luns cannot be used with metaluns

Updating the host information

Right click host > properties > storage > update

Only available when the host agent is installed or it is an ESX/ESXi host

Info includes – SCSI devices, physical device address, FS on device, storage system the device belongs to

VNX protection suite – snapview, snapsure, recoverpoint/SE CDP

Clones track changes of source lun in the fracture log

Clones of snaps can be managed by the unisphere gui, navisec cli, admsnap

Snapviw copies use approximately 20% of the size of the source lun

When a clone is fractured the clone tracks the data changes to the source lun via the fracture logs

Snapshot session can be up to 64 characters long

Snapshot chuncks are set to 64k (128) blocks

Snapshot pre-reqs

· Source luns must be bound

· Assign the snaps to a storage group

· Enable data access control

· Reserved luns must be bound

Best praxtice to have clone luns and source luns on separate disks

Reserved lun recommendations

Total size of the lun/numberof source luns

Reserved lun size = 10% average source lun size

Create twice a smany reserved luns as source luns

Example 10Gb,20Gb, 30Gb , 100Gb

160\4 = 40Gb

Make each reserve lun 4Gb in size (40Gb x 10%)

Make 8 reserved luns (twice as many as source luns)

Snapshot sessions are persistant

Persistant – will survive failures and trespasses (SP reboot or failure storage system failure or reboot

Consistant –

· preserves point in time across luns

· available on per session basis

· counts as one of eight sessions per lun

snapshot roll back

source lun available for reads and writes

copy happens in back ground

user should flush host buffers using admsnap

source device should be offline

snap taken off ine

VNX SAN copy – source lun can remain online during copy

A clone lun needs to be exactly the same size as source lun

When removing clones it cannot be in an active synchronous or revers sync process

Following can’t be cloned

· hot spare luns

· clone luns

· snapshot luns

· private luns

clone sync rules

· sync from source to clone lun

· clone private luns used for incremental sync

· fracture logs saved persistently on disk

· two luns of at least 1Gb each

· host access

· source access i/o at all times/ clone no i/o during sync

clone reverse sync rules

· source must be taken off line prior to reverse sync

· during re-sync all writes to clone then sync’d to source

clone reverse sync – protected restore

· protected

o host > source writes not mirrored to clone

o configure via individual clone property

· non-protected

o host > source writes mirrored to clone

o records source writes in the clone private lun for subsequent syncs

for protected check the use the protected restore in the add clone dialog box

“instant restore” 8 clones per source lun

Consistant fracture when you fracture more than one clone at the same time in order to preserve the point in time restable copy. Across the set of clones

Consistant operations from unisphere or NavisecCLI

You cannot perform a consistant fracture between different storage systems

If there was a failure on any of the clones, the consistant fracture fails on all clones. If any clone within the group was fractured prior to the failure, the software re-syncs that clone.

Clone consistant failure

· source lun for each clone has to be unique

o cannot fracture multiple clones that belong to the same source

o cannot consistently fracture if one clone is already fractured

· clone appear as administravely fractured (one fractured)

failures affect all clones in group

mirrorview /s and mirroview/a cannot mirror a clone

snapsure

· snapshot point in time view of data

· savVol stores original data blocks to preserve the point in time view and holds modified blocks in case a writable snap is used

· bitmap – identifies changed data blocks in the production file system

· blockmap – records the location of the data blocks in the savVol

· baseline snap – read-only snap from which a writable snap can be created

a new block map is crerated for each snap and the bitmap gets cleared

bitmap and block map saved to disk but paged in data mover memory up to 1Gb of RAM is allocated for snapsure

a bitmap consumes 1 bit for every 8k blocks in the pfs

a blockmap consumes 8 bytes for every 8k block in the snapshots saVVol

1Mib sys memory for every Gib in SavVol

For systems with less than 4Gb 512Mb is allocated to data movers for blockmap storage

Server_systat –blockmap shows dm mem consumed by blockmaps

SavVol sizes usually 20% of the size of the PFS

Manual savVol creation

· manually create VNX meta volume

· specify that metavolume when creating the snap

· volume sizing limits apply

a full savVol invalidates all snaps using it i.e all snaps become inactive.

A savVol autogrow extends in 20Gb increments

Extension triggered by HWM (default HWM 90%)

savVol will not exceed 20% of the total NAS storage (this can be changed in the parameter file /nas/sys/nas_param

if not enough disk space will overwrite the oldest snapshot

to disable savVol auto extend set HWM to 0%

snapsure will overwrite the oldest snap

auditing and extending are important to prevent unwanted data loss

snapsure limitations

· max 96 snaps per pfs + 16 writable snaps total 112

· do not schedule snaps on the hour as this will conflict with the VNX OE for file’s database backup

FSCK – sanity check the PFS, without taking t offline, by running fsck against a writable snap

Writable snaps

· can be mounted and exported as read only

· share savVol with read only snaps

· add write capabilities to both local and remote

writable snaps must be deleted before the baseline snap

writable snaps cannot be refreshed (snap schedule)

baseline read only snap can have 1 writable snap no snap from a writable snap

per PFS

· 16 writable snaps

· 96 read only snaps

· Total 112 snaps per pfs 96 + 16

Scheduled refresh on read only snap will fail if writable snap exists

Warning displayed when a writable snap created on a scheduled r/o snap

Total number of fs per cabinet 4096

Max number of mounted fs per data mover 2048

Writable nas limits

Not supported with:

· VNX replicator

· File system extension with writable snap

· VTLU’s?

· Timefinder f/s

Writable snaps cannot be accessed through CVFS (Checkpoint virtual filesystem) gives users read-only access to snaps. Saves admins from having to run backups for small restore requests.

Name of hidden directory *.ckpt

Name customisation via the param file

CVFS names can only be changed when remounting the snap via CLI only

Shadow copy clients via the Microsoft volume shadow copy server

Nas_ckpt_schedule

Refreshing snapshots

When you refresh a snapshot, VNX snapsure deletes the snapshot and creates a new one recycling savVol space and maintains fs name, ID and mount size.

Scheduled snaps that are baselines of writable snaps will not refresh until the writable snap is deleted.

Checkpoint refresh of a writable snap is not supported

Refreshing a snap requires freezing the snap. If an NFS client tries to access the snap during a refresh, the system continuously tries to connect indefinitely when the snap thaws the snap is re-mounted. If a CIFs client attempts to access a snap during a refresh depending on the app, or if the system freezes for more than 45 seconds, the app may drop the link and the share may need to be re-mapped.

PFS needs to be paused to delete snaps in order by doing this it could have an impact on performance.

When you delete a snap that is not the oldest snapsure compacts the snap and merges the block map entries to the next oldest snap before the delete completes.

RecoverPoint

CDP (Continuous Data Protection)

CRR (Continuous Remote Replication)

CLR (Concurrent local & Remote Replication CRR & CDP for same volume)

Recover Point Appliance

Use	Min Size	Max Size
Repository Volume	3Gb	3Gb
Journal Pools	5Gb	10Tb
Replication Volumes	X	2Tb(minus 512)

One journal for each consistency group is needed. They hold snaps of data to be replicated. Usually 20% of the size of the data replicated.

Host Based Splitter (kdriver)

Installed on the host above the scsi multi pathing driver

Kutils is installed with the driver

I. App write i/o

II. f/s

III. vol mgr

IV. kdriver splits writes

V. powerpath

VI. scsi driver

VII. hba’s

when splitting is performed by the switch a standalone version of the kutils must be installed on the host

array based splitter

· requires recoverpoint/SE enabler

· lower cost

· increased scale

· supports mixed splitter environment

· RPA storage group needed

· RPA needs to be registered with the array

· Array and RPA ports zoned with each other

· Splitter runs in the SP’s

Multi cluster array based splitter

· Supports up to 4 recover point/SE clusters

· LUN is dedicated to single recoverpoint/SE cluster

· Max of 2048 LUNs across all clusters

· Recoverpoint/SE 3.3 or higher

· Array splitter 3.3 or higher

CDP uses

Write is split to RPA when received acknowledge sent. This ack is received by host splitter and held until ack is received from host when both acks are received the ack is sent to the host. When RPA has acknowledged write it moves data into the local journal.

A consistency group consists of one or more replication sets

Distributed consistency groups span four sets of RPA’s

5Gb for regular consistency groups

20Gb for distributed consistency groups (4 x regular consistency groups)

DCG’s support gen 3 or 4 max of 8 distributed consistency groups per cluster.

CG’s – Replicating

I. The CG is enabled

II. Splitter is replicating to the RPA’s

III. RPA’s are replicating to remote journal

IV. If image is access is disabled (default state)

V. The journal also distributes snapshots to the replica storage

Marking

I. The CG is enabled

II. Splitter replicating to RPA’s

III. RPA’s unable to replicate to journal

IV. When link is up will carry on syncing

Disabled

· Splitter does not split writes to RPA’s is caused by CG being set to diable or by a disaster

Data marking and Transfer

· Full sweep – when group is enabled for 1^st time

· Volume sweep – sub of a full sweep

· Marking mode – when new rep pair is added to a CG

Fisrt initialization

Initialization

Marking mode

· Short resync

o Delta between source and target is small

o Caused by high load, WAN outage

· Long resync

o Delta too long for journal

o Snap segmented and transferred

o Segment distributed before next segment transferred

Image access enable mode (logged access)

Allow_long_resync

System pane, traffic pane, nav pane, component pane

Configure luns as read only ( used by service providers)

To create a read only lun

Storagegroup –addhlu

Can only be done in the CLI

NavisecCLI storagegroup –list shows LUNs that are read writable or read only

Storagegroup –list –readonly

To make the lun read writable remove from storage group

Storagegroup –removehlu

Add lun back in using –addhlu without read only

AVM best practices

Slice volumes by default (slice box enabled by default)

Thin box enabled by default

Pool striping

First step divide luns into thin and thick groups it will then try and stripe 5 dvols together of the same size, same data services, and in an SP balanced manner. If 5 can’t be found AVM will try 4, 3 then 2 dvols to stripe. If thick luns/dvols can’t be found it will use thin luns/dvols

Pool concatenation

If AVM cannot find enough thin or thick luns/dvols it will then try and concentanate enough thick luns to meet the size requirement if not enough thick luns then it will try and concenate enough thin luns and if that fails it will concenate thin and thick luns/dvols and if that fails it fails

Course notes

Best practice for spares is 1 every 30 disks

Exam tip read all white papers on block and file

FE errors ultraflex i/o modules

M errors

Be errors

Access logics handles initiator records

Linux supports 16 initiator ports (exam question)

Windows supports 32 initiator ports

HLU – Host lun id

LALU – Array lun id

Vault disks first 4 (flare code, firmware, DART)

7500 – SPE storage processor Enclosure

DPE Disk Processor Enclosure

RAID 3 modified on vault drives helps for large sequential reads/writes

VNX is active/passive

(Exam question) whats the command to give user root access idroot_user

Nas/bin/addrootidtouser root_user

(Exam question) looking at alerts it uses SSL for event auditing and alerts

Kerberos time authenticaton – 5 mins

(Exam Question) nas/bin/celerra is where the logs are kept

(Exam Question) x.509 certificates on VNX

Certificate assigned to SP ip’s change sp IP need new certificate

Max 16 disks in a raid group

RAID 1/0 bus balancing

RAID 1 is used to configure FAST CACHE

RAID 3 bit parity

RAID 4,5,6 block parity + dispersed

Read cache

Write cahce

System cache

Applications will read sequentially if written well for performance

Max 255-256 luns on raid groups

RAID group best practice

Use 70% of total IOPs

1000 IOPs needed

200 IOPs per disk

Use 7 disks

7 x 200 = 1400 IOPs

First most available space will choose vault drives when creating luns using the gui

(Exam Question) default for creating luns is storage pools not RAID groups

Storage group –list –readonly will show read only luns

Max disks in a pool are all attached minus the 4 vault disks

Elab navigator shows compatibility matrix

Data written in 8 byte chunks

· 2 x write stamp

· 2 x time stamp

· 2 x parity info

· 2 x CRC

Up to 2Tb of FAST CACHE on 7500

Control station gets connected to slot0

Block LDAP master pushes info into slaves

File LDAP has to be configured on each slave

For file logging into control station you need to be a nasadmin to run config commands

Write cache

Forced flush above 90%

Idle flushing below the low watermark default 70%

Using water makr algorithm

(Exam Question) NavisecCLI –h x.x.x.x port –list –hba | more

(Exam Question) scli sansurfer determines HBA’s

Linux lun rescanning needs i/o quiescing

Linux 2.4 kernel is disruptive best to reboot

Linux 2.6 kernel has dynamic lun scanning

Most effective way to find new luns in linux is to perform a reboot

Cat /proc/scsi/scsi shows hba/path/lun id’s

Use the above to verify luns

(Exam Question) where to verify luns /proc/scsi

Iscsiadm –m node to show target node

Verify luns with powerpath

Powermt display dev=all

ALUA – ascymetric lun access

Powerpath failover needs array failover mode 4 for alua to work

Update al hosts to verify luns in unisphere

(Exam Question) what can oversubscribe CPU cycles? SMP can do this

VSI-emc API’s (distributed hierarchical storage manager)

DSHM needs to be configured for cloning, dedupe to work

(Exam Question) certificate can not be MD5 signed to by pass certificate checking

Each run time luns

Which is the lun id or does not change in run time name=first 9 naa.xxxxxxxxxxx

Use server utility to check high availability verification (HAVT)

System reports > high availability reports

(Exam question) when using shared storage each esx erver needs the same lun id for tasks like vmotion or strage vmotion

(Exam Question) what are some file and block features/extensions

Storage –dedupe

File – compression

Metaluns

Element size multiplier size = 4

Upto 16 metaluns striped if you want more i/o takes time. Concenate if space is needed quickly

Lun migration limited to 44mbps

Uaed to change RAID type of lun to move so that you have lun with more space

When doing NDU with metaluns it can delete data

In storage pools using bitmap to locate where luns data blocks are

Compression

Needs to de-compress to read or write the dat

(Exam Question) compress at 64k chunks

(Exam Question) FAST VP moves data in 1Gb chunks

Sparse luns support will keep polling for new luns if lun id’s do not match up

If data is hit 3 times it will stay in the cache you can have up to 2Tb of cache configured using FAST CACHE and RAID 1 configured EFD’s

Lun shrink on w28 uses DISKRAID.exe

File

Uxfs – is the filesystem which will hold CIFs and NFS shares

DART – Digital Access Real Time

Uxfs holds uids and guids and sids

NTP is needed for Kerberos authentication

For datmover to talk externally they need a default gateway configured

To telnet to control station you need to be nasadmin

FSN – fail safe network

Server_sysconfig – command to list pci devices

Server_ifconfig – command to show IP address info

Ipv6 only for mgmt puposes configure ivp6 at cs properties or CLI

Default gateway needs to be configured before CIFs server is added to the domain

CLI command to configure DNS

Server_dns server –o stop

Default protocaol for DNS is udp (because its faster)

NTP Status – server_date server time svc stats ntp

Server_devconfig used to configure devices

Luns chopped up into disk volumes dvols

File system (upto 16Tb) > meta volume (at least 2mb) > slice volume (needed for snapsure) > stripe volume > disk volume > luns > raid groups

Can only extend file system pools with system defined pools not user defined pools

(Exam question) two pool types for file are – system defined/user defined

Two types of pools for block are – virtual pools/raid groups

Can have raw file systems that you can provision straight to the host

Uxfs is like wafl

DART islike ontap

Auto extend buffer size is 3% below the HWM

Nested file system can export a group of single file systems as one

Use vsi to attach nfs storage

Usermapper only for CIfs to map UIDs and GIDs

Sec map holds UIDs and GIDs

(exam Question) you should not touch starting point for UIDs and GIDs

(Exam Question) what does the value start at 32768

One primary user mapper

Virtual data movers only for CIFs

Need to configure CIFs on each data mover

See if CIFs has joined domain CIFs server porperties, background taks s for file in system

Global Share don’t choose DM

Local share choose DM

Supports SMB2

Overview

· Uxfs – native filesystem for NFS, CIFs

· Usermapper – 3 roles primary, secondary, client

· Storage pool profiles for file – system defined/user defined

· Two storage profiles – volume/storage

· Sliced volumes – used for snapsure

· DART – Data Access Real Time

Virtual Data Movers are for CIFs only

Root file system for data mover – 256Mb

Root file system for virtual data mover – 128Mbvirtual data movers need to be read only in order to be migrated

Normal user quotas

Tree quotas

Quotas configured on the parents

Soft limit quotas will warn hard quota limit will deny writes if gone beyond quota allocation

Tree quota overrides normal user quota

8191 max tree quotas per file system can’t have tree quota in a tree quota i.e cannot be nested

(Exam Question) by default on “file size” policy quotas need when you have a new directory is set in the file system path

Home directory supports CIFs use NT security only

Domain:username:/path

CIFs home directory structure as above

In etc folder you can add a wild card so that you do not have to input each user individually

Domain:*:/path

Map home drive command

Net user esele/domain/homedir:llceldmz/home

Recommended to use CAVA sizing tool to determine how many CAVA servers

Exclude fles are configured on viruschecker.conf

CAVA only available for CIfs

Filtering done CIFs share level

Filter files stored in \.filefilter

Upto 8 devices in an etherchannel

2,8 up to 12 in a link aggregation LACP

(Exam Question) what is the definition of VLAN trunks 802.1q

Cge=copper gigabit Ethernet

Fge=fibre gigabit Ethernet

Fxg=optical 10 gigabit Ethernet

Network failures will not cause data movers to failover to the standby DM

Cross stack etherchannell is between two seoerate switches

2 types of monitoring centralised and distributed

Monitoring host cannot be connected to storage or be in any storage groups

VNX alerts

· Warning

· Critical

· Informational

· Error

VNX alert Responses

· Email

· SNMP, Syslog

· Custom Application

· Paging, SMS

30,000 hex code errors and alerts

Unisphere host agent needs to be installed on the monitoring host

Use server utility to check if esx has high availability

(Exam Question) Fault icons

· F – Fault

· X – inaccessible

· U – unmanaged

· T – Changing State

· ? – Unsupported

2 weeks retention for file statistics up to 26 weeks

Audit events in etc/audit/auditruler

Main record types

· Syscall

· PATH

· CWD

· USER_XXX

· FS_WATCH

Each control station sync’d every 180 seconds

(Exam Question) where are the audit logs located? /celera/audit

User events from MMC windows event logs

Snapview for block

Up to 8 snapshot sessions per LUN

4 snapshot sessions per RPL when over 4 sessions it will need to use the next RPL

Need protection enabler for snapshots

Admsnap to schedule snaps

Before starting a snap you will need to sync servers memory buffers using the admsnap utility

Overview

· LACP – 6 active links

· CAVA – Anti Virus

· Snapshots – COFW 8 active

· VDM’s

· Quota’s – normal quotas/tree quotas

· CIFs auditing enable through MMC snap in

· VNX auditing enabled by default

· Network Failover, FSN, LACP, Etherchannel, Cross-Stack etherchannel

· Monitoring – centralised/Distributed

· File extension monitoring – noext

Cloning both luns need to be the same raid type the same type of disk or you will get performance hits

All volumes/File System’s use savVol to save snapshots

96 read onlysnapshots per file system

16 writable snapshots per file system

Snapshots will have a block map

Deleting a snapshot requires you to delete the writable snap first then the read only snaps

Bit map will get deleted along with the snapshot

A new bit map is created for each new snapshot

You cannot snap a writable snapshot

CVFS naming convention

Yyyy_mm_dd_hh_mm_ss

Snaps cannot be scheduled at beginning of the hour as this conflicts with the VNX OE database backup