FalconStor: Simplifying Data Protection for the Enterprise - PowerPoint PPT Presentation

1 / 57
About This Presentation
Title:

FalconStor: Simplifying Data Protection for the Enterprise

Description:

VTL services (Reduce Backup Cost and Complexity) ... VTL Backup. 0110110. Repository. Seconds. Weeks. Weeks. Months. Years. Remote Sites ... – PowerPoint PPT presentation

Number of Views:361
Avg rating:3.0/5.0
Slides: 58
Provided by: Eliz194
Category:

less

Transcript and Presenter's Notes

Title: FalconStor: Simplifying Data Protection for the Enterprise


1
FalconStor Simplifying Data Protection for the
Enterprise
2006 Users Conference
Christopher L Poelker VP Enterprise
Solutions FalconStor Software
FalconStor Making the complex simple
2
Common Information Technology Goals
  • Institute a Services Oriented Architecture for
    compute, network and storage resources while
    leveraging standards and best practices (ITIL)
    to
  • Reduce overall IT costs (CAPEX/OPEX)
  • Reduce risk
  • Improve ROI
  • Improve service quality
  • Leverage more business value out of IT
  • Enable greater business productivity and
    flexibility
  • Improve business objectives for RTO and RPO at
    the lowest possible cost with the best possible
    return

3
Common Customer Requirements
  • Solution must integrate with currently installed
    SAN and backup solutions
  • Solution must improve backup and recovery speed
  • Reduce tape media requirements
  • Eliminate tape shipping
  • Reduce hardware costs
  • Reduce footprint
  • Integrate with existing policies and procedures
  • Reduce storage requirements for retention
  • Improve RTO and RPO objectives
  • Provide centralized management

4
FalconStors Portfolio of Enabling Solutions
FalconStor Vision Virtualized Data Services with
Continuous Protection
  • Heterogeneous Storage Virtualization (Gain
    Operational Efficiency)
  • Multi-Protocol Storage Provisioning and Pooling
    (FC, IB, iSCSI, iSER, NAS)
  • CDP services for all data (Unified Data
    Protection)
  • CDR services for all data (Cost Effective
    Disaster Recovery)
  • VTL services (Reduce Backup Cost and Complexity)
  • Data reduction (Cost Effective Archive with User
    Based Recovery)
  • Regulatory Compliance (SEC17a4, HIPPA, etc.. WORM
    compliance)
  • Remote to Core services (Consolidate Backup,
    Protect all data)
  • Desktop/Laptop services (Continuous Protection
    with Bare Metal Recovery)
  • WAN optimization and Encryption (MicroScanRUDP
    Reduces WAN Costs, Encryption Reduces Risk)
  • PrimeVault DR Service

5
FalconStor Unified Data Services
Complex Business Applications
FalconStor Technology
Heterogeneous Provisioning
Virtual Data Services
Heterogeneous Storage Pooling
Unified under a single Platform
SAN, iSCSI, NAS, IB, Desktop, Remote Office
Application Quality of Service, Capacity on Demand
iSCSI, FCP, IB, iSER, CIFS, NFS and GRID Storage
Simplicity and Value
IPStor
Host Transparent Data Migration
Data Encryption, Compression
Core Data Protection Services
Redundant Data Elimination
Virtual Tape Backup
WAN Services
Continuous Data Protection and Replication
Remote Office Data Protection
Desktop, Laptop Data Protection (Bare Metal
Restore)
Off-Site Disaster Recovery and Hosting
(PrimeVault)
FalconStor Control of Resources
6
Traditional SAN Infrastructure
7
Future-Proof Datacenter Paradigm
8
Application Focused Recovery
Web Servers
App Servers
DB Servers
DAS
DAS
DAS
DAS
DAS
DAS
CDP
Fabric
3 Tier App
Database Primary
DB
Log
CDR
Business Logic Modular
App
Consistency Group
Single Image Recovery
9
Future Proof Data Protection
Fabric
  • TimeMark application consistency
  • Guaranteed write order fidelity
  • Point and Click DR roll reversal
  • Encrypted data
  • Data Reduction
  • Transparent Data Migration and Tech refresh

Any Vendor
Any Vendor
10
Recovery LifeCycle Management
FalconStor is the only company today that enables
unified data protection and recovery during the
entire lifecycle of data
Single Instance Archive X1
Any Application
Remote Sites
0110110
Continuous DR Replication
Repository
Off-Frame Copy
Tape Export
VTL Backup
Any Storage
Consistent Tape recovery
Consistent Virtual Tape recovery
Consistent off-frame recovery
Consistent DR recovery
CDP journal 23459 23458
Jan
200pm
Mon
200pm
Feb
300pm
Tues
300pm
Mar
400pm
Wed
400pm
Apr
500pm
Thurs
500pm
Seconds
Weeks
Weeks
Months
Years
11
Information LifeCycle Management
FalconStor is the only company today that enables
transparent data protection and recovery spanning
the entire lifecycle of data
Single Instance Archive X1
Tier 1 Production
Tier 2 Recovery
Tier 3 DR
Tier 4 Archive
Storage Class Enterprise
Storage Class Bulk/Tape
Repository
Storage Class Modular FC
Storage Class Modular SATA
Unified Host Transparent Data Movement Via IPStor
Engine
Consistent Tape recovery
Consistent Virtual Tape recovery
Consistent off-frame recovery
Consistent DR recovery
CDP journal 23459 23458
Jan
200pm
Mon
200pm
Primary
Feb
300pm
Tues
300pm
Mar
400pm
Wed
400pm
Apr
500pm
Thurs
500pm
Seconds
Weeks
Weeks
Months
Years
12
TimeMark Storage Efficiency Reducing Storage
Costs
TimeMark snapshot is extremely efficient
  • Standard block size is a function of the size of
    the file system
  • They have different bitmap granularities
  • Measure in Kbytes (4K, 8K,64K,OCFS128KB)

Appl Server
KB
KB
CDP
KB
KB
KB
KB
KB
KB
KB
  • Microscan turned on increases granularity to a
    sector level
  • Measured in 512 bytes

3
2
1
Snapshot Resource Area
B
B
B
B
Full Volume
B
B
B
B
B
  • Less storage required for each retained TimeMark
  • This is also a bandwidth metric Less data
    transferred

13
MicroScan Data Replication Efficiency
  • Minimum amount of data being sent over a WAN link
    for each write
  • FalconStor is
  • 64 times more efficient than EMC
  • 32 time more efficient than HDS
  • 8 times better than NetApp or Kashya
  • EMC 32KB
  • HDS 16KB
  • NetApp 4KB
  • Kashya determined by file system
  • FalconStor 512bytes

FalconStor dramatically reduces WAN bandwidth
requirements
14
RUDP Performance Test 45Mb
Improving WAN Efficiency
Peak 49Mb/s
Peak 4Mb/s
FTP 200MB file 45Mb Bandwidth 4,000 miles 1-way
(NY to Sweden) 80 ms Round Trip Delay .12 packet
loss
Native 635 Seconds
RUDP 50 Seconds
15
MicroScan Efficiency Actual Case Study
Reduce WAN and Storage Costs via MicroScan
16
Storage and Replication ROI
T3 connection 5.5925 MB/s 10 overhead 5.03
MB/s,(18GBH) 7 500-14k/month OC1 connection
6.48 MB/s 10 overhead 5.83 MB/s,
9k-18k/month OC3 connection 19.44 MB/s 10
overhead 17.49 MB/s, 10k-30k/month OC12
connection 77.76 MB/s 10 overhead 69.984
MB/s, 100k-300k/month OC48 connection 306.104
MB/s 10 overhead 275.4936 MB/s,
300k-500k/month
Data WAN link Normal Replication Time With MicroScan WAN needed without MicroScan to replicate in 10.7 hours MicroScan Annual WAN Cost Savings
1.2TB T3 68 hours 10.7 hours 2 x OC3 600,000
Note Assumes 84 MicroScan Reduction
17
DiskSafe and FileSafe Backup (Blocks or Files)
Days or weeks of on-disk recovery using
dramatically less storage than other backup
solutions Always a Logically Full
Backup User Enabled Self-service
Restore MicroScan RDE minimizes
bandwidth Data is encrypted on transport Bare
Metal Recovery (DiskSafe)
Windows Servers/Desktop/Notebook
SmallOffice
FileSafe
DiskSafe
LocalDisk P
LocalDisk Q
WAN
LAN (iSCSI)
Central Site
1st Backup Couple of Hours From then on
Couple of Minutes Only Deltas are transmitted
and stored Always a Synthetic FULL
Backup Advanced data hashing/de-duplication (Save
WAN costs and Storage costs) Can replace
existing backup software Better functionality
than Avamar, Kashya, Data Domain, Symantec
PureDisk
Disk P Files
Disk Q Blocks
Centralize protection for remote locations and
desktop data
80 Less Storage
18
User Enabled Restore of any File Version
User enabled recovery from repository No backup
application required Immediate access to multiple
versions of data Searchable index
19
Remote Office Consolidation
WAN Throttling and Tuning Auto-resume on WAN
outage/recovery Complete Write Order
Fidelity Integrated Redundant Data
Elimination Built-in Remote service/Callhome/SNMP
Bare Metal Recovery
Windows Servers/Desktop/Notebook
FileSafe
DiskSafe
DiskSafe
LocalDisk P
LocalDisk Q
Local Disk R
LAN (iSCSI)
All Recovery Done Locally Including BMR All Data
Replicated for DR 80 Less WAN 80 Less
Storage Encryption Included RUDP WAN Optimization
Included
Disk P Files
Disk Q Blocks
Disk R Blocks
Better functionality than Riverbed, Avamar or
Data Domain
20
Example Corporate Data Protection
Small Office
Small Office
Small Office
Small Office
Small Office
r-hub Site2
r-hub Site3
r-hub Site1
Continuous Replication
r-hub Site4
r-hub Site5
Small Office
SmallOffice
SmallOffice
Small Office
Small Office
21
Solution Example
Primary 1
Primary 2
22
Actual Savings
  • 95 better Recovery Time
  • Tape was four to 12 hours
  • Eliminated Backup Windows
  • 18MB/s to gt80MB/s per LUN (Approx 3.5TB per Hour)
  • 99 improvement in RPO
  • Near zero data loss vs. 12-24 hours
  • Use of Space Efficient Copies, (not full copies)
  • Use of Microscan, (RDE), to save 85 of storage
    Space
  • Total Tier II Disk 2.5 TB vs. 20 TB of Tier I
    storage
  • 85 saving on bandwidth required for DR
  • 8 MB/sec, (100 mbps) vs. 60 MB/sec, (Full GigE)
  • Use of Tier II Storage educed solution costs
  • Eliminated backup windows
  • Applied new DR capabilities for ALL applications

23
IPStor Directions
24
IPStor 2006
  • IPStor v5.1 Constellation Released June 2006
  • Continuous Data Protection Journal
  • Allows users to select more granular points in
    time for recovery
  • System CrossMirror
  • High availability for appliances without
    requiring shared (FC) storage
  • 4Gb Fibre Channel support

25
IPStor 2007
  • IPStor v5.x Olympic Release Early 2007
  • InfiniBand target support
  • High-speed connectivity using iSER
  • Centralized Management
  • DiskSafe, FileSafe, DynaPath, Snapshot Agents
  • Support for LUNs larger than 2TB
  • Encryption at rest
  • Multipoint replication
  • Multi-hop replication (Local Sync to Remote
    Async)
  • Multi-Node IPStor N-WAY

26
Continuous Data Protection Journal
How It Works
  • Continuous
  • Host writes to primary data disk
  • IPStor or DiskSafe duplicates write to CDP
    journal area
  • Journal
  • Journal tracks data changes before they are
    committed
  • TimeMarks are recorded in journal as special
    entries
  • Replica
  • A full, independent copy of the primary data
  • Updated from journal as needed (based on time or
    when journal fills)
  • TimeMarks
  • Used for long-term data protection
  • Snapshot Agents for point-in-time recovery of
    databases, e-mail, file systems
  • LUN consistency groups
  • Recovery
  • TimeMark(s) can recover data to fixed times
  • Journal can recover data written after TimeMark(s)

Journal
PrimaryData
Replica
TimeMark 3
TimeMark 2
TimeMark 1
27
CDP Recovery to last uSecond
28
Traditional FalconStor Virtualization and Data
Protection Methodologies
Host Based Out Of Band
Host Based Side Band
Host Based LVM
Fabric Based Out of Band
Appliance Based In Band
H
H
H
H
H
DAS
DAS
A
B
A
B
A
B
A
B
A
B
IPStor
IPStor
IPStor
IPStor
IPStor
P
S
P
S
P
S
P
S
P
S
Native Host LVM.
ASM, SANTap
IPStor SED IPStor Virtual Disk
DiskSafe FCP
DiskSafe iSCSI
H Host ABSwitches P Primary S Secondary
Notes Host based facilitates protection of root
volumes Host Based can be Block, File,
or Both In-Band facilitates transparent
ILM, Tech Refresh, COD, and Virtual Provisioning
29
New Virtualization and Data Protection
Architecture
N-WAY
CDP Triplet
H
H
IPStor
CDP
CDP
IPStor
IPStor
A
B
IPStor
IPStor
IPStor
A
B
IPStor
IPStor
IPStor
IPStor
IPStor
P
S
IPStor
P
S
IPStor
Failover Cluster Triplet
N-WAY Enterprise Instance
  • Best practice methodology based on throughput and
    connectivity requirements
  • N-Way connectivity supports Up to 32 paths per
    LUN with ZERO failover time

30
N-Way Virtualization with CDP Detail
Each IPStor treated as a path Each IPStor can
have many 4Gbit Connection Paths 2PB Storage
Supported N-Way managed as a single instance
Host
  • Up to 32 Paths Per LUN
  • Host Transparent Failover
  • Online Upgrades/Maintenance
  • Dynamic Load Balancing
  • Massive Scalability

A
B
CDP
CDR
IPStor
IPStor
IPStor
IPStor
IPStor
IPStor
IPStor
IPStor
P
S
P
R
31
N-Way with CDP Mirror Replica and TM
Host
Each IPStor configured Identically. Service
Enable the same Physical LUN, assign to Host, and
Mirror to IPStor-CDP. Supports DynaPath AND
PowerPath in Active-Active mode.
P
M
SED
M
SED
SED
M
FC Mirror
FC Mirror
FC Mirror
TM1
P
TM2
M

TMn
32
CDP Fast (Delta) Recovery
  1. Primary LUN P is also assigned to IPStor-CDP.
  2. Establish Mirror from R1 to P without full Sync.
  3. Perform TimeMark or CDP Rollback to repair R1 and
    P
  4. Remove/Suspend Mirror

Primary Site
IPStor CDP
NearlineServer
FS Mirror
Primary
TimeMarks
Mirror
P
CDP Cache
33
Online Banking Example (East Coast)
IPStor Console
34
Enhanced DR and Archive Environment
Legato
Encrypted Tapes
CDR
CDP
HyperTrack
VTL
SATA
SATA
35
Oracle RAC Environment
DL380 Apps
RX4640 RAC
Recovery Server RMAN
Active
IPStor Cluster
Active
8FC
8FC
4FC
4FC
CDP Journal
CDP/CDR
EVA3000
36
Oracle RAC Disaster Recovery and Archive
DL380 Apps
RX4640 RAC
Tapes Encrypted on Export from VTL
Active
4GBFC
CDR
HyperTrack
IPStor
VTL
SATA
SATA
37
IPStor v5.x System CrossMirror
  • Configuration
  • 2 matching IPStor appliances
  • DAS/internal storage (50 utilized)
  • Interlinked using iSCSI to share storage
  • Every drive is duplicated
  • Normal operations
  • LUNs visible from target ports on one IPStor
    appliance
  • Writes to LUN are copied to drive in partner
    before returning to host
  • Failover procedure
  • System detects failures of partner and takes over
    if needed
  • Survivor assumes operation for failed partner
    LUN access moved to survivor
  • Survivor tracks changes while partner is down
  • Benefits
  • Benefits of mirroring and failover
  • No shared (FC) storage
  • Low Cost!

Host
Switch
A
A
IPStor Servers
O
C
38
Self Healing Compute and Storage
IBRIX LUSTRE PLATFORM POLYSERVE Oracle RAC Etc.
iSER
iSER
Cross System Mirror
39
Example State Government iSCSI stretched cluster
City A
City B
Sync Mirror
Local Disk Access
Local Disk Access
Gigabit Ethernet
iSCSI
iSCSI
iSCSI presentation of FC disk to enterprise
appliances across Gig-E link allows transparent
failover between sites with NO SAN!
40
Microsoft Exchange Stretched Cluster
500GB
500GB
Host LUNS
Cluster Link
Site A
Site B
Bi-Directional Mirror
FC Distance
41
Polyserve and FalconStor Matrix GeoCluster
Active Production Site
Active Recovery Site
Continuous Data Replication
42
IPStor GRID Solutions
43
Why Infiniband?
44
Commodity iSER Target or Router
Voltaire/Topspin InfiniBand Switch
Performance
900 MB/s
Target
730 MB/s
Initiator
400 MB/s
Initiator
FC
iSER (1 Target)
iSER (gt1 Target)
Delivering unprecedented cost/performance,
leveraging on industry standard components
45
IHM Institute for Health MetricsFC Free Data
Center
Data from Hospitals
Voltaire Grid Director ISR 9096
Native InfiniBand Storage with FalconStor IPStor
WAN Router
Local Users
InfiniBand VTL
ApplicationServers
Oracle 10g RAC Cluster
46
Virtual Tape
47
What is FalconStor VirtualTape Library?
  • Disk storage array emulates tape drive or library
  • Provides high-speed data transfer
  • Leverages FalconStor VTL services to simplify
    tape management
  • VTLs HA architecture guarantees the
    backup/restore operation wont be interrupted

Application Servers with 3rd Party Backup Software
TSM
Emulates a tape library
Backup Media Servers
Fibre Channel or iSCSI
VirtualTape Library Appliance
RemoteReplication
  • Import/ Export from/to physical tape without
    impact to servers and with no backup window
  • Exported tapes can be injected into a non VTL
    infrastructure
  • Tapes can be replicated and encrypted

Import Export
FC-attached RAID Storage (FC and/or ATA)
Any Primary Storage
48
VirtualTape Library Releases
  • VTL 5 Rockwell Release Q1 2007
  • Multi-node group
  • Single Instance Repository
  • Backup data contains 80 typical redundancy
    between full backups
  • Reduces VTL disk storage needs
  • Reduction done outside backup window to avoid
    impact on application servers
  • Tape Stacking
  • Export many virtual tapes to a single physical
    tape
  • Allows VTL to make most efficient use of
    available physical media
  • Preserves investment in older tape equipment
  • Allows users to transition to newer equipment
    gradually
  • Tape Shredding
  • Uses DoD 5220.22-M secure erase standard to
    expunge data
  • 4 Gb QLogic HBA support (Active/Active

49
MultiNode Group
  • Linear Scalability
  • Up to 8 nodes (mixture of single nodes and HA
    pairs)
  • Performance
  • 8 800 MB/s 6.4 GB/s 23 TB/h throughput
  • Capacity
  • 8 64 512 tape libraries
  • 8 512 4096 tape drives
  • 8 32K 256K virtual tape cartridges
  • Unified Management

50
FalconStor Single Instance Repository
51
FalconStor VTL with Single Instance Repository
Servers
Data Coalescence Time Based Inter-Nodal Intra-Noda
l Sub-Block
Backup Software Agnostic
Client Enabled Restores
User Based Recovery Multiple versions of files
can be viewed and restored independent of backup
application
gt80 Reduction
Backup Software Independent
5001
52
Shadow Virtual Tape
  • Single Instance scan
  • The file data is extracted and added to the
    repository.
  • File data is replaced with links to the extracted
    data.

Original Data
VolumeHeader
SessionHeader
DirectoryMetadata
File Data
File Data
File Data
FileMetadata
FileMetadata
FileMetadata
53
Shadow Virtual Tape
  • After Single Instance Scan
  • The resulting virtual tapes contain only the
    backup metadata with the links to the repository
    file data entries.
  • The links are the keys to retrieve the data when
    needed.
  • These links are used by either the backup
    application or FileSafe client to recover data
    from the repository

VolumeHeader
SessionHeader
DirectoryMetadata
FileMetadata
FileMetadata
FileMetadata
FileLink
FileLink
FileLink
Links
File Data
File Data
File Data
Single Instance Repository
54
Existing D gt D gtT Backup
Retention 1 year Total Media 300TB
Tape Archives (12 Monthly)
240TB
20TB
20TB
20TB
Tape Copy
Primary
D to D Copy
Note At Zero Change rate for simplicity
55
VTL with SIR
Retention 1 year Total Media 72.4TB
gt400 Savings in year 1 media costs
20TB
20.4TB
24TB
Primary
Tape Archives
Simplify D to T IS D to D copy
Re-purpose original D copy as VTL
4TB
VTL
SIR
56
Example Virtual Tape with SIR
Value Leverage existing storage for automated
data protection and archiving Savings in this
case 500,000 less per year in media
costs 90,000 savings per month in WAN costs
57
Summary
58
The FalconStor value model
Enable Business Flexibility
Decrease Operational Costs
Increase Service Levels
  • Vendor Choice
  • Time-to-market
  • Application impacts on
  • Provisioning
  • Protection
  • Security
  • Recovery
  • Staff utilization
  • Storage utilization
  • Server utilization
  • Backup costs
  • Network costs
  • Media Costs
  • Storage Costs
  • Better Availability
  • Faster recovery
  • Higher SLA
  • Impacts on...
  • Customer Satisfaction
  • Customer Retention
  • Zero Lost business

59
Thank You!
Write a Comment
User Comments (0)
About PowerShow.com