Generative Artificial Intelligence: How generative AI works.pdf
It delivery 2016 v5
1. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph
IT modern
delivery
It’s all about the IT execution!
2. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 2
What is IT delivery?
3. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 3
Delivery challenges: Tsunami of new technologies
4. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 4
Delivery challenges: the different IT types require
different delivery characteristics and level of effort
5. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 5
The STKeye - three pillars of Delivery:
Development and related SW tools
PaaS, IPaaS, CaaS, microservices , Docker, agile , DBMS, Devops, Dev Tools
Core Infrastructure (compute, storage, network)
Storage, network, server that delivers IaaS and are the foundation of PaaS and CaaS in the
private or public cloud enable Devops
Other delivery domains
Public cloud Integration for SaaS, external API, System monitoring (ESM), ITSM , End user
computing , IT procurement, Testing tools
Deliver
Development
Deliver IT
Execution
Other delivery
domains
6. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 6
Fast and
efficient
Great DX
(developer
experience)
Cloud ready
(enabling
scaling, utilizing
modern cloud
functionality etc.)
Portable,
managed
(traceable, etc.)
Production is
updated
constantly
Keep up with
new
technologies
Based on Open
Source
components
Good SW Quality
(QA)
Secured
software
architecture
and
development
Development : the vision is clear
7. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 7
Modern development principals
Rest
Use rest as standard API
1
Web Development
Do not use client server unless
a must
2
Automatic Tests
In all layers to enable Devops
also unit tests and TDD test
driven development
3
No technology debt
Try to have supported
technologies
4
Agile software development
Use Agile as default development
methodology
5
Enhance and not replace
?Enahance transactions” means
“co-existence”). Nosql might help
here.
6
Devops
Devops (both code and
infrastructure)
7
Microservices
Microservices (but use ESB) , fault
tolerance service (smart proxy)
and enable polyglot development
8
Application logs
Application logs preferably “event
sourcing”- writing each change in
application status to log
9
8. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 8
Microservices
Source: http://martinfowler.com/
9. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 9
MicroservicesHow big is your team?
10. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 10
APaaS – application platform as a service
Developer’s heaven
On public cloud Private and
public cloud
11. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 11
Development and Deploy stack – before containers
Devops
tools(Jenkins,
Nolio ,
Urbancode)
Physical HW
Infra
automation
(puppet, chef,
ansible)
APaaS
(CloudFoundry,
Openshift)
Virtualization
layer (Software
Defined)
Cloud
Management
Platform (vRealize,
Openstack, Azure
stack)
12. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 12
This might change with containers
13. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 13
Source: files.meetup.com/11185112/Docker-Meetup-
jan-2015-Final.ppt
The basic challenge
“binaries libraries” complexity
14. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 14
Source: files.meetup.com/11185112/Docker-Meetup-
jan-2015-Final.ppt
The Matrix from hell….
15. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 15
Virtual Machine Vs. Containers
OS
HW
16. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 16
What are Containers and Docker ?
• Linux Containers (LXC) is an operating-system-level virtualization method for
running multiple isolated Linux systems (containers) on a single control host (LXC
host).
• Docker is an open-source project that automates the deployment of applications
inside software containers, by providing an additional layer of abstraction and
automation of operating-system-level virtualization on Linux. (Wikipedia)
17. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 17
Why it Works: Separation of Concerns……
Source: files.meetup.com/11185112/Docker-Meetup-
jan-2015-Final.ppt
18. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 18
Devops
tools(Jenkins,
Nolio ,
Urbancode)
Physical HW
Infra
automation
(puppet, chef,
ansible)
APaaS
(CloudFoundry,
Openshift)
Virtualization
layer (Software
Defined)
Cloud
Management
Platform
(vRealize,
Openstack)
All layers will support containers
19. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 19
Container schedulers and orchestration
20. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 20
Devops
tools(Jenkins,
Nolio ,
Urbancode)
Physical HW
Infra
automation
(puppet, chef,
ansible)
APaaS
(CloudFoundry,
Openshift)
Virtualization
layer (Software
Defined)
Cloud
Management
Platform
(vRealize,
Openstack)
What will be the role of container
schedulers?
Container schedulers (Mezos,
kubernetes, swarm)
21. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 21
Things are still moving: Microsoft and VMWARE
VMWARE: ‘Instant Clone’ also
known as VM Fork or Project
Fargo
22. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph
Deploy modern
SW architecture
Development
architecture and
technologies
Advanced SW
development and
architecture will enable
efficient, portable,
secure, stable and
manageable
development as long as
providing great DX
(developer experience) to
the developers
Cloud native
Microservices
Polyglot agile development
Open source technologies is
majority
Evaluate agile
Start to deploy
modern languages
Experiment with
cloud
development
Only
maintaining
legacy SW
New
development
is outsourced
23. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 23
STKI Recommendations: container technologies
and APaaS
Do now
Be aware
Start to select and implement new technologies for
development and deployment
Processes, organizations and management
commitment is more important than tools
selection
Looks like APaaS is more mature than container
schedulers
24. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 24
The STKeye - three pillars of Delivery:
Development and related SW tools
PaaS, IPaaS, CaaS, microservices , Docker, agile , DBMS, Devops, Dev Tools
Core Infrastructure (compute, storage, network)
Storage, network, server that delivers IaaS and are the foundation of PaaS and CaaS in the
private or public cloud enable Devops
Other delivery domains
Public cloud Integration for SaaS, external API, System monitoring (ESM), ITSM , End user
computing , IT procurement, Testing tools
Deliver
Development
Deliver IT
Execution
Other delivery
domains
25. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 25
IT Execution – the vision is clear
26. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 26
IT Execution – the vision is clear
ERP
Security
deep packet
inspection
Storage
deduplication
Machine
learning batch
program Online pricing
optimization
27. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 27
IT Execution – the vision is clear
Security
deep packet
inspection
Storage
deduplication
Machine
learning batch
program
Constant
Deploy of
code and
infra
Burst into
cloud
ERP
Balance
with other
private
clouds
Online pricing
optimization
28. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 28
15
Standard cheap
Servers
No centralized
storage
Cloud = elastic &
self healing,
templates,
orchestration
workflow, etc.
Delivered
constantly to
production
(Devops)
Bursting to
public cloud and
balanced
between private
cloud locations
Network HW
fast but stupid
Security based
on micro
segmentation
All
“virtualized”
and “software
defined”
IT execution (running the IT) : the vision is clear
Adequate
organization
structure
29. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 29
Compute
Wistron Server
30. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 30
STKI Recommendations: Compute
Do now
Currently main stream servers are the safest
choice
For new type of workload cheaper servers
should be examined
31. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 31
Software Defined Storage solutions
HPE
32. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 32
All Flash Array: “AFA born” vs. “AFA migrants”
• Can put flash logics (“even writes per cell”) at basic OS level and therefore
use less expensive flash
• Will use (all the time) inline dedup and compression globally (not per
LUNRaid group)
• Kaminario
• EMC XtremIO
• SolidFire (NetApp)
• HP 3PAR 8450
• IBM All Flash (TMS)
• EMC VNX-F; VMAX-All Flash
• NetApp All Flash FAS
• HDS VSP-F
• HP 3PAR
• IBM V7000
• Dell Flash Storage
1000 erases per block (TLC)
33. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 33
Object storage
Distributed
No limits (name spaces, scale)
Metadata
34. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 34
STKI Recommendations: Storage
Do now
Be aware
Implement SDS for none production and later
for none critical storage
Centralized storage is here to stay for the short-mid
term
Implement initial object storage especially for
large NAS environments
Backup to public or community cloud if regulation
allows
35. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 35
User: LindaUser: James
HP UC&C SDN
Application
HP VAN SDN
Controller
Active
Directory,
Exchange &
SharePoint
.
Lync
SDN API
HP Server HP Server
Traffic
Generator
Traffic
Generator
HP OpenFlow Switch
HP OpenFlow Switch HP OpenFlow Switch
Ringing…
Networking: SDN business applications
36. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 36
SDN: Network Micro-segmentation for better security
Source:
https://www.ku
ppingercole.com
/blog/kuppinger
/beyond-
datacenter-
micro-
segmentation
37. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 37
Network – SDN players
38. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 38
STKI Recommendations: Network DC
Do now
Be aware
Put explicitly what is the need for SDN and see if SDN
(and which) is the answer
Professional decision and not “Network vs.
System vs. Storage” political war
Large organizations – play with SDN even if there is
no immediate reason
Security is domain with high potential for SDN
deployment
39. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 39
Compute, network & storage: together or
apart?
compute network
storage
40. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 40
Modern platforms (converged infrastructure and
cloud platforms) dimensions:
HW vs. SW1
Infrastructure blocks
(vBlock) vs. standard
servers
(hyperconverged)
2
Based on traditional
network HW vs.
SDN based
3
Level of openstack
support
4
VMware based vs.
other hypervisor
5
Basic execution (compute,
network, storage) vs. cloud
capabilities (templates,
workflow, elasticity, ...)
6
41. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 41
Cloud platform and Converge infrastructure – so
many options…
“serving Devops, built easily from “menu” by automation”
42. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph
“Running the
systems” with
compute,
network and
storage
IT
Execution
advanced IT execution
will enable elastic and
efficient DC operations
deployed via devops
state of mind secured via
micro segmentation and
aligned to business
needs
Deploy Hybrid cloud for
production with CSB as
mediator managed by
unified “IT Execution”
team
Deploy first cloud
applications
(DMZ, dev-test-
train, etc.)
managed by
coordinated
system teams
Relay on traditional
technologies with
separate system,
network DC and
storage teams
43. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 43
Cloud Platform and Converged Infrastructure
Do now
Be aware
Currently, for short term “cloud
ready” project VMWARE is the
natural choice
Containers might change everything…
Hyperconverged solutions are becoming ready for
prime time
Goodbye “Server-Storage-Network DC teams”. Long
live “IT execution” team
44. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 44
The STKeye - three pillars of Delivery:
Development and related SW tools
PaaS, IPaaS, CaaS, microservices , Docker, agile , DBMS, Devops, Dev Tools
Core Infrastructure (compute, storage, network)
Storage, network, server that delivers IaaS and are the foundation of PaaS and CaaS in the
private or public cloud enable Devops
Other delivery domains
Public cloud Integration for SaaS, external API, System monitoring (ESM), ITSM , End user
computing , IT procurement, Testing tools
Deliver
Development
Deliver IT
Execution
Other delivery
domains
45. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 45
Enterprise System Management: the march towards
the logs
46. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 46
SBC and VDI is mature
47. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 47
Testing Automation enable Devops
48. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph
Enable modern
delivery in all
domains
“Other IT
delivery”
Modern IT delivery will
enable smooth and
efficient operation with
good tracking of events
and change management
processes based on high
quality code with
superiour end user
experience
All tests are automated
and enable “continues
delivery” with advanced
log based BSM
Experiment with
log based BSM
while deploying
VDI for niche
projects
Based on
traditional BSM
technologies
with little test
automation
49. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 49
Final word
50. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 50
Where are my projects located on the
maturity graph?
R
51. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 51
STKI Recommendations: Why delivery does
projects fail?
Be awareLack of management commitment
Unrealistic expectationsbudgettime
Team A will is using the deliverables but Team B is
managing the project (example for cloud ready project :
Infrastructure vs. developers)
Deliverables are not 100% clear and constant change
in requirements
52. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph 52
Summary- Lets ride the tsunami wave!
But focus on where you want to get!!
53. Pini Cohen’s work Copyright@2016. Do not remove source or attribution from any slide or graph
That’s it.
Thank you!
53