Grid v. Cloud
I was pointed to a document that looks at the differences between grids and clouds:
I was pointed to a document that looks at the differences between grids and clouds:
Previous strategic plan required projects to demonstrate how they fit w/ strategic plan to get money
Previous plan had IT as a silo. New plan: IT isn't a silo. IT as part of other silos..
WiFi RFP completed: upgrade over the summer
Implementing MPLS w/ PCI as the first target
Isolating and protecting devices on the network
- automatically detect some types of devices and create an appropriate environment for them
- 802.1x on wired and wireless
- non-802.1x devices - proxy w/ MAC address
- web based registration for MAC addresses
Important to have diagnostics to know what happened
security principles
- Security is Everyone's Responsibility
- Security is part of the development life cycle
- Security is Asset Management
Risk assessment process
- letter of engagement
- conduct assessment
- draft report
- communicate findings
- reassess
Virtual Roundtable
It's all about the data. Doesn't matter what compute power you have.. the value is in the data
Data Governance - fairly well understood (relative to other areas of data management)
Major issues
[Taking notes on challenges with individual scenarios in a spreadsheet]
scenario 1 - finding data in a data warehouse.
scenario 2 - information is not easily consumable.
scenario 3 - privilege management. access based on role, not on username
"role based access control continues to be the access control method of the future"
whole issue of RBAC is hard, needs to be thought out. [not sure this is the most important in data mgmt?]
but, externalizing the management of privileges separate from the app makes things better
scenario 4 - electronic lectures and talks. metadata storage
CAESAR project - classify unstructured data
scenario 5 - data going onto web2.0 - PKI & signing
scenario 6 - medical data
scenario 7 - ediscovery
Though most of my day to day 'blogging' occurs on Twitter, I wanted to mention here that I'm going to be leaving Duke. My last day will be June 26th.
At the end of June, Rebecca and I will be relocating from North Carolina to the northern Virginia area. I have accepted a position at Amazon.com; specifically Amazon Web Services (AWS). I will be joining a small software team developing new products under the AWS umbrella.
As the saying goes, this move comes with mixed emotions. I have and continue to enjoy the challenges and opportunities at Duke. Duke's IT organization has forward-thinking management and an experienced staff. It's difficult for me to say goodbye, yet I am excited by the new opportunities that await me.
While I'm thinking about it, let me point out some interesting work that is occurring on AWS:
JBoss on EC2
Building an EA Team - acquired / re-purposed staff with architecture experience
Focused on future state
Plan strategy iteratively
EA Team - process arch, enterprise data arch, operations, app integration, security, network, web, deputy CIO
Think about end user experiences
Developed architectural principles - based on TOGAF framework
- Business principles, data principles, application development principles, technology principles
Developed IT Guiding Principles for centralized & decentralized staff -- "How we'll function"
Discovery - lack of rigorous requirements management -- solutions misaligned with problems
Starting to use COBIT, evaluating ITIL.. TOGAF/PRINCE2 for planning & organization, IIBA/PMI/DAMA for Acquire & Implement, Six Sigma to monitor & evaluate
Silos
No linkages between silos
Passive change control
"show me an ROI" for enterprise architecture
Blogging from ITANA Face 2 Face 2008
8 System domains, 70 applications (originally 7 applications.. too many interdependencies)
Process
Developed model for how apps are clustered -- e.g. groups, relationships ("org chart" ?)
Conceptual approach to framing roadmap
Strong focus on spending wisely; communicating spending decisions to raise awareness of what they have
Planning to develop common policy/architecture
Focus on having the architectural vision apply throughout central and decentral departments
Current status: agreement to implement
More Info
Side item - UCDavis is moving students to gmail
Product lifecycle - Researching > Current/deploy > Stop deploying > Phase out
Focus on specific products, services -- e.g. desktop computers and printers
Architectural Council - 19 folks w/ architect in the title (how does that work?)
Architecture Review Board - 9 folks
EA Team - chief architect + 2 architects
Procurement department constrains purchases based on PIM
19 of 105 IT staff are considered architects (broad-banded, 4 titles, 1 title is architect)
Fewer people have the total perspective on how it all works
IT Ecosystem tool (requires login)
Simple ontology - verbs and nouns to describe "stuff" (servers, database, "liveware" (users, groups), platforms, apps, storage, network, etc), relationships
Questions
Relationships - support, modify, part-of, hosts, xfers-data-to, other
Folks copy the tool and use it locally to plan infrastructure, etc
Network folks planning to start integrating data center infrastructure in the system.. don't know what impact they'll have with network moves -- [does someone have the visibility on both network and system to help the understanding?]
Data center folks planning to use to coordinate moves/adds/changes there
Some concept of DataCharacteristics -- e.g. SSNs stored somewhere
No current way to show current v. proposed -- e.g. facilitating change management
Not really looking at using this as a CMDB in the ITIL sense.. [why not? what should be the system of record for this data?]
Can follow dependency chains - textual and graphical reporting
Number of things that depend on a particular service -- SAN is highest [where is DNS?]
Tested 4 infrastructures - Cisco IOS, Cisco LWAPP, Meru, Aruba
w/ multiple laptops:
macbook pro (3x3 atheros)
macbook air (2x2 broadcom)
systemax (2x2 intel 4965)
ibm (4965 3x3)
HP (something)
results
- seems to be 40mhz tests
- aruba had greatest throughput across the board (surprise?), 150mbps
- meru had very odd results (some really low)
- cisco ios mid-range
- cisco lwapp 2nd in most (up to 100mbps or so)
- aruba numbers seem similar to numbers we saw from LWAPP in august
talking about one channel in 2.4ghz, making the case for 5ghz.. not mentioning 20mhz channels??
"Can that access point run using a single 802.3af power injector over a single 100 meter cable?"
- saying yes we can
- not mentioning that it runs with reduced functionality
- second generation atheros chipset
Chuck Bartel @ CMU -
4.1M square feet being deployed
wireless andrew .. 8100+ regd devices, 5000+ simultaneous active
targeting -70dBm RSSI in all spaces
Greater focus on software than hardware
Bluetooth, 802.11n are the big winners..
NFC, Wireless HD, ZWave all players .. zwave v. zigbee
software controlled becomes software-defined radios becomes cognitive radios
software-defined: multiple protocols, modulation techniques.. all digital. now available
cognitive radios require regulatory changes .. 4 years out
approximately 3yr lag between HDD and flash memory for similar price per gb
going to multi-core in mobile devices.. able to turn off cores that you aren't using and power them up for multimedia, etc
OLEDs coming to displays.. LightBlue Optics .. micro-projector
location is important.. GPS in 40% of the handsets in a few years. doesn't work indoors, of course.
mapping WiFi access points based on MAC address and using that to determine location.
another idea: take a picture of where you are, compare against a massive database of photos
power is a big problem, no big changes
opportunity to break mobile phones into pieces .. e.g. display as a wristwatch, antennas/radios in your pocket, etc. connected wirelessly
usability - final frontier
good designers are scarce
no standards for gestures, etc
interesting new paradigms
- vibration as feedback mechanism
- shoogle - shake to test battery
- bluetooth bonding - shake together
- multi-device virtual environment .. wave your hand around to control mouse
good products/services: visceral impact (5 second response), functional (does what you want), reflective (goals, values)
adoption drivers
- single tech supplier
- uses established standards
- easy
- uses existing behavior
- simple value proposition
- compelling value
- visceral appeal
- social value / cool
- visible success
Tis the season.. Tuition Hikes at Duke
... and many others. But CMU crossing the $50K threshold is a bit disconcerting.
So let's see.. if CMU continues rising 6% per year, on average, as it has for the last 8 years, in 20 years tuition will be a cool $125,000/yr. Obviously financial aid offsets, but it seems to me that this trend can't continue as it has.
I'm here in Chicago, where it's cold and rainy, with a promise of snow. I'm giving a spiel tomorrow morning on 802.11n.
Unified Communications ...
- intro slide: please sit in the last 2 rows if you are taking notes on your laptop. 20% of the room has laptops out.. also, "battery operated only for safety reasons". e.g. we didn't want to pay union rates to have power outlets. let's see how long 46% lasts me. :(
work.. from desk worker, to mobile worker, to tomorrow's semi-mobile worker. mobility as a driver moreso than cost
unified comm is: PBX, email, voice mail, converged conferencing, IM (+presence & status)
learning from contact center, collaboration, business process integration, mobility, etc
continuum
1:1, many to many
internal/external
discrete/embedded
persistent/temporary
fixed/mobile
message/real-time
business case for UC? hard to write it across the board.. focus on the obvious areas and grow from there
thinking about UC within organization, also across organizations..
** need to raise the questions of how this works between organizations.. early in the maturation process, do we have the right standards in place? shib, etc
different levels of support based on device and connectivity ... wow, how will we survive?!?!
[tuning out]
fixed mobile convergence.. all the rage. not going to save $ right away
standard quadrant..
conferencing: interwise (now AT&T Connect) .. eh
adomo: low cost .. AD & Exchange. eh.
SIP is coming. accept SIP.