The Official Salesforce Blog
The Official Salesforce Blog - On-Demand Vision
-
IT Opportunities Rise to the Cloud
Oct 19, 2009Cloud computing's naysayers encourage misplaced fears that IT careers are threatened by cloud adoption. Don't believe it. The economic argument for clouds as career opportunities is clear; the need for insightful expertise to realize the cloud's full potential is becoming widely recognized.
But when it comes to rebutting cynical FUD, theory is nowhere near as strong as practice—and the practice of IT is clearly moving cloudwards, as witness this current job posting:
Job Title Project Manager
Location Arlington, VA
Position Id 5788Job Description
Resolvit is looking for a full time employee to work as a Project Manager at our client site in Rosslyn, VA. The successful candidate will utilize their practical experience to manage the development and deployment of ad hoc enhancements and changes to the recently deployed enterprise wide CRM tool (Salesforce.com). A major conversion from Seibel was done and there are many new opportunities for a project manager to get control of, prioritize, and manage to success.Duties:
· Develop detailed project plans to include timelines, resource allocation estimates, budgetary analysis, risk factor identification and risk factor mitigation strategies.
· Gathering requirements, specifications and best practice documentation from multiple stakeholders on how to enhance and roll out new features for Salesforce.com CRM tool
· Perform gap analysis and produce strategies and tactical plans to enhance the customer's experience
· Create and manage the task schedule of deliverables as well as monitor the budget for each new program
· Working with groups of 3-10 users to facilitate acceptance and delivery of new CRM tool modifications
· Provide progress reports to management, escalate issues as needed, identify and minimize risk factors, assess project changes and their impact to schedule and budget
· Reviewing the quality and consistency of all changes and enhancements throughout the development processPlease notice two things: the valuable contributions of an IT professional are still right up there on the list; the low-value, career-stagnating parts of the job are not.
To me, this looks like the beginning of a revival of the opportunity—and even, perhaps, the joy—of pursuing an IT career.
-
A Cloudy View of the DJIA
Oct 16, 2009Reasonable people may cordially disagree on the merits, as a measure, of the Dow Jones Industrial Average—which is in the news, this week, even more than usual after its return (however transient) to the zone above 10,000.
Yes, this threshold is arbitrary; worse, it's surprisingly slippery, thanks to the frequent changes in membership of the group of 30 stocks whose prices go into the calculation.
Regardless, let's take advantage of the huge mind-share of the DJIA, and use it as a lens through which to view the past and speculate on the future.
Calculating the DJIA is trivial on any given day, but it poses tricky questions if you want play "what if?" with alternate histories. I got into this game back in November 1999, when Microsoft and Intel were brought into the Group of 30 for the first time in place of Chevron and Union Carbide. In a PC Week column a few days later, I expressed displeasure with Dow Jones Corp. for chasing (as it seemed to me) the glitter and glitz of the tech-heavy NASDAQ Composite. (The latter index would peak, as we doubtless all remember, only four months later at a height that it has never since approached.)
I've since noted, for example in an eWEEK column in March 2005 (so far as I know, not available on line) that subsequent struggles of the DJIA would have been far less labored if Microsoft and Intel had not been holding it back. If we look back to December 2003, for example, when the DJIA crossed above 10,000 for the third time in its history, we can ask ourselves: what would have happened since then if Microsoft and Intel had been re-replaced by the stocks that were voted off the island four years earlier?
Others may approach this calculation differently, but as I compute this, today's DJIA would have topped 10,120 if MSFT and INTC had been sent back to the stock yard.
But wait, there's more. What if MSFT and INTC were replaced, not by two solid though unexciting petrochemical players, but instead by two exemplars of the cloud? Say, Amazon and salesforce.com? The latter did not go public until 2004, so this calculation has even more ways to do it than the one before—but as I can best run the numbers, a DJIA that had turned its eyes cloudward at its previous 10,000 breakthrough would today be nearing 10,760. We might not be dancing in the streets, but some of the parties that were held this week could have been more pleasant summer barbeques.
For investors, perhaps this mind game is a useful reminder that indicators aren't the same as measures. Sure, the DJIA is based on objective measures, but the continual re-composition of the DJIA massages those measures into well-informed opinions. There's nothing wrong with that, unless you misunderstand what the number means.
For everyone else, I offer this as an exercise in thinking about what will drive our economy forward during the decade to come. The "Wintel" model of thick-client computing had a good long run, but its day is nearly done. "We expect Windows and Office to be empty shells" a decade from now, says Toan Tran, an equity research analyst at Chicago's Morningstar Inc.
There's more to the market than simple numbers, but numbers are the basis of all real knowledge. Don't get caught in a shell game.
-
The Cloud Without the Airfare
Oct 9, 2009I spend a lot of time on airplanes, en route to give conference keynotes and other such briefings, most of which are centered on the question of "What's this cloud computing thing and what will it do for me?"
On one rare recent occasion, not only did I get to talk about this with just a short drive up the freeway to Los Angeles, but our partner colleagues at American Data Company also did a super job of combining the slideware with the speech.
This talk starts with the big picture of the IT economy, lays out both the diversity of cloud computing offerings and also their common tenets, and ends with an update on our Sales Cloud, Service Cloud and Custom Cloud perspectives.
Hot Topics in Enterprise Cloud Computing (PaaS and SaaS) from American Data Company on Vimeo.
Best of all, unlike the real thing, a video version of my talk has a "pause" function, and doesn't expect hot coffee before it starts. PDF of the slides available for download as well, for those who want to play our game at home.
-
Positive Feedback Has Impact
Oct 9, 2009People often use the phrase, "positive feedback," to mean the mere act of praising good (or even mediocre) work. In technical settings, the phrase has a more precise and rather different meaning: positive feedback makes small changes self-reinforcing, with an everyday example being the howls and screeches that come from a public-address system when the microphone gets too close to the speaker.
The power of positive feedback came to mind earlier this morning, when I joined a standing-room-only crowd at Northrop Grumman's Space Park complex in Redondo Beach for the 4:30am'ish end of the LCROSS mission: a pair of lunar impacts, the first of a spent upper-stage rocket and the second of an instrumented probe designed to sniff the debris plume for water.
As we neared the moment of impact, several people involved in the mission were sharing their lessons learned from this fast-track project that achieved under-budget, ahead-of-schedule results—despite aggressive targets that some thought unlikely to be met. "People said we couldn't do something this small, this fast," was the recollection of one of the engineers. "When we announced that the spacecraft was ready for delivery early, it ricocheted all the way up to NASA headquarters."
I found myself thinking of positive feedback. When you know you're trying to do something in two years, rather than ten years, you feel more motivated to work for an extra twenty minutes to get something done today—instead of planning to do it before lunch tomorrow. The extra effort feels as if it's worthwhile.
I wonder if this may be one of the reasons that cloud computing projects get done more quickly than anyone expects, or even believes to be possible. An extra twenty minutes, before morning coffee or during a working lunch or squeezed in before you go home, can make a contribution that's really tangible when you're working in a high-leverage framework like Force.com. Every action yields perceptible progress.
This is pretty much the polar opposite of the joke (at least, I think it's meant as a joke) that asks "What's a man-year of effort at [insert big slow organization name here]? That's 700 people saying they can get it done before lunchtime." Massive, legacy-burdened projects always seem to take longer than anyone thought, and throwing more people at them just makes them take even longer.
People ask me how Force.com achieves fivefold acceleration of development, compared to platforms like Java or .Net: I tell them it's not just the programming model. It's the whole environment in which reusing data, and multiplying the value of existing processes by linking them easily with additional resources, gives every increment of effort more impact than it's ever had before. Positive feedback indeed.
-
It's Coming from Inside the House!
Oct 8, 2009When people raise questions of data protection in clouds, I've often noted the ample research available on internal versus external threats—as tabulated, for example, by the GAO in that agency's analysis of sources of vulnerability in U.S. government IT (figure at right). This is explored in more detail by a white paper available by link from one of my earlier blog posts.
If you're not segregating duties, if you're not auditing the access privileges that you grant with reference to the processes you want to enable, then it's really quite irrelevant where the information is being stored.
Expanding on this key understanding, a new report from InformationWeek's Dark Reading site now puts the situation in even plainer terms with the headline, Databases' Most Serious Vulnerability: Authorized Users.
"There are five common factors," says the summary of the report on the Dark Reading site, "that lead to the compromise of database information":
- ignorance
- poor password management
- rampant account sharing
- unfettered access to data
- excessive portability of data
Many of these sources of risk are directly and dramatically reduced by practices that are easier to implement in cloud services than they are in traditional client-server environments. Salesforce.com systems are excellent examples.
For example, every individual salesforce.com subscriber is associated with a unique set of login credentials, which even the administrator has no direct way of knowing: a password reset operation sends required information and activation links directly to the subscriber, meaning that any actions to access or modify data can be unambiguously associated with a specific person.
Access to information can be controlled with unsurpassed precision.
Excessive portability of data is another important source of risk to consider. The path of least resistance in client-server settings is for data to expand to fill the space available: to wind up downloaded onto desktops, backed up onto thumb drives, attached to emails, and in general copied in too many places and shared via too many unsecured and ungovernable channels.
In a Force.com environment, it's far more natural for people to share links to shared content libraries, rather than making N (or multi-N) copies of the data for N users. This library model makes it far more likely that updates, redactions, or altered access policies will have the desired (or even mandated) effect of making sure that people only see what's correct—and also, not insignificantly, what's genuinely needed to do their jobs.
"Experts say that many users who work with databases simply don't understand the sensitivity—or the value—of the data they work with, and therefore become casual in their security practices," warns the Dark Reading summary. That's not going to change any time soon, and not without incurring expenses that most organizations don't want to face.
At least we can build applications in an environment that makes insecure behavior less convenient than disciplined data management.
-
More is So Very Not Better
Oct 7, 2009It's staggering to read the descriptions of the multi-city-block, polar-icecap-melting monuments to code bloat that are being built to run 20th-century software in a poor imitation of 21st-century cloud computing. When I got my undergraduate degree in civil engineering, I hoped someday to build nuclear power plants: I never imagined that a Chicago server farm would soon be "the first point of consumption for the nuclear-fueled Elmhurst power grid, initially taking 30 megawatts of power, enough power to supply 20,000 U.S. homes, with plans to take 30 megawatts more."
There's a better way. Instead of running N copies of a traditional, general-purpose operating system for N separate users of that operating system, a multi-tenant kernel can do the job with a tiny fraction of that amount of code on a tiny fraction of that amount of hardware. As noted by one attendee at an event in New York this March, "all of salesforce.com runs on only about 1,000 servers. And that is mirrored, so it is really only 500. Think about that for a minute. Salesforce has more than 55,000 enterprise customers, 1.5 million individual subscribers, 30 million lines of third-party code, and hundreds of terabytes of data all running on 1,000 machines." (Today, it's more like 63,000 customers and more than 80 million lines of Force.com code, but it's the basic ratios rather than the absolute numbers that matter here.)
Compared to the massive old model, running applications in the new multi-tenant model is like putting people in business-class seats on an A380 from L.A. to Singapore—versus putting the same people, in their cars, on a ferry boat that goes about one-tenth as fast. Sure, a turbine-powered hydrofoil ship is a beautiful piece of engineering, just like that data center in Chicago—but isn't the destination more important than the machine that gets you there?
And doesn't it make sense to get there faster, more efficiently, at lower cost? By not defining the problem in terms that make you keep on carrying the burdens of past practice?
The differential savings of new versus old are only going to increase. As Ed Sperling asks over on Forbes.com, what happens when carbon taxes start to do a better job of fully internalizing costs that today are not fully visible in the prices we pay for electric power?
Putting thousands of copies of a 20th-century operating system in one big place, and calling it a cloud, looks to me an awful lot like putting wings and jet engines on the Titanic and calling it an A380. Burn enough fuel, and you might even make it fly—but the day of the flying boat came and went with startling speed. Other notions may pass just as quickly, no matter how clever the engineering...or how impressive the monuments that are built to the bad idea.
-
The Cloud Should Be the Center, not the Edge
Oct 2, 2009Someone asked me yesterday how to convert a document from a proprietary word-processing format to a PDF file. "All the things I can find on line are downloads that mess with your printer settings," he said. I told him that OpenOffice 3.0 does a lot of nice new things with PDF export (and also document import, including Office 2007 formats); I suggested, though, that by far the easiest solution was to upload the file into Google Docs, then download it back to his machine as a PDF.
This is more than just a useful tip for dealing with the world's unfortunately long list of legacy and proprietary formats. (It's especially sad that so much time gets spent addressing these problems for content that's mostly simple documents and alphanumeric tables. If you want to send someone some formatted text, why not send it as RTF or HTML—instead of sending it in a vendor- and release-specific format that's weighed down with irrelevant exotic capabilities?)
What's more important is to start thinking in terms of "the cloud as first resort." Don't ask, "what can I download to do this on my PC?" Ask rather, "what cloud service will take what I have and deliver what I want, where I want?"
Instead of thinking, "how do I send this to someone as a PDF?", shouldn't the question be something more along the lines of,
- How do I make this content accessible to someone without assuming that he has some specific vendor's application installed on his machine?
- How do I make this accessible to a whole bunch of people without being limited to some common subset of content formatting capabilities?
- How do I share this with an open-ended audience of people without sending a separate copy to everyone who wants to see it?
And then you start to think about authoring documents on line, or at least publishing them on line, in ways that let others contribute—while still giving you visibility into who's done what, and control of who can do what.
The cloud should not be the mere connective tissue between pockets of local capability. It should not be the last thing that you use, after you've done everything else that you can possibly do with local facilities. The cloud should not be at the edge of your IT map.
Rather, the growing range of powerful and convenient services in the cloud should be the center of your thinking about what to do, in the quickest and most economical way—because that center is getting bigger, and richer, and more attractive in every way, every day.
-
What the Cloud Entails
Sep 23, 2009You can't put the cloud in a box, let alone a pigeonhole. People who ought to know better continue to pretend that cloud applications aren't ready for full-spectrum enterprise workloads, but enterprise developers do know better—and they're writing the apps today.
You can see the distribution, or at least get an idea of its outlines, in the results of a study (PDF) just released by CustomerSat, a unit of MarketTools Inc., from a sample of more than 1,100 companies using Force.com. In striking agreement with previous, totally separate studies by Nucleus Research and Galorath Inc., CustomerSat concluded that "on average, it’s 5.1 times faster to develop custom applications with Force.com compared to traditional on-premise development platforms."
To some extent, that's old news; what's new, at least to me, in the CustomerSat results is the information now available on the breadth of applications being built. To quote the CustomerSat report,
The most popular applications were for reporting and analytics, project management, and contract management applications. Next, respondents reported building more transaction-heavy applications, including quote and order management, billing, and commission/compensation management applications. Finally, long tail applications ranging from compliance solutions to legal management applications followed.
That phrase, "long tail," is probably known to most readers of this blog: we're nearing the fifth anniversary, somewhat to my surprise, of Chris Anderson's coinage of this label in Wired. The basic idea is that most of the potential value in a market may well be in items with tiny individual market share, representing the far extreme range of a frequency distribution. The problem in a physical marketplace is that artificial scarcities, like shelf space in bookstores, encourage vendors to stock only the items with the largest individual sales.
Enterprise application software has always been a "long tail" market, because return on investment was high enough to compel costly custom development as the price of being a competitive company. We've traditionally wound up with a marketplace in which a handful of monstrously complex products were individually modified, site by site, to meet their customers' needs.
Multi-tenant kernels and metadata-based customization, as found in Force.com, make it possible to serve the long tail with economy of scale. That's a whole new game. Let's play.
-
Adding Up to Agreement
Sep 22, 2009Yesterday, I called attention to a poll being done by The Motley Fool on the magnitude of the cloud computing opportunity. Votes have been piling up since then, and the numbers are more than a little interesting.
Right up top, a nice majority of respondents unreservedly agree that cloud computing is "the biggest market opportunity right now" – but some of the other possible responses also belong in the same column of cloud computing endorsements.
For example, some voters agree that the cloud is the major opportunity, but think that cloud providers' share price levels are too high. The market will decide, but clearly nothing else looks better right now. Others feel that the cloud is not the last word – but then again, does it need to be?
When you sort the votes into just two buckets, "pro" and "con", only 18% of voters consider the cloud either over-hyped or under-secured. I'll argue cloud security any day. As for hype, I remember lots of things that have had that accusation thrown at them – and thrown it right back. (“Well informed people know it is impossible to transmit the voice over wires and that were it possible to do so, the thing would be of no practical value” – The Boston Post, 1865)
The real question, it seems to me, is this: of all the faults that one might ascribe to the cloud, is there any single item on that list that won't be substantially addressed within the next three years? Of all the faults that burden older models of IT, is there any single item on that list that won't still be with us?
Are there any other questions?
-
Only a Service "Just Works"
Sep 22, 2009I've been a Macintosh user (sometimes exclusively so) for 25 years, so I'm allowed to use Mac TV ads as examples of a pernicious fantasy: the myth that any machine "just works".
Apple's TV ads get a lot of mileage from the notion that Macs "just work", while other popular thick-client platforms putatively plague their users with "a ton of headaches". In real life, there's almost never such a thing as a machine that you buy, and own, and use, and never need to debug or maintain.
My own favorite counterexample to the Mac "just works" myth is the Spotlight search utility in the OS X operating system. To be 100% completely fair, Spotlight's just great when it works. When it doesn't, though, there's a tedious and counter-intuitive process for rebuilding the search index structures: you add your target storage location to a Privacy list in Spotlight Preferences to make Spotlight forget whatever it thinks it knows about the contents, then remove that item from the list to let Spotlight get to know the contents all over again.
Sometimes, it takes a system restart to make the Privacy list pay attention; when the process works, it's up to you to figure out when it's done. Imagine telling your parent (or your boss) over the phone, "Open the Activity Monitor, pull down the options list at top right and click it to view all processes, then wait for mds and mdimport to stop taking turns doing CPR on the file system." Sounds like a headache to me.
Could things be better? Sure. In a robust operating system, you can always add one more level of monitoring. If I start a Spotlight search, and nothing happens in the next fifteen seconds, I could conceivably get a pop-up box that says "Spotlight seems to be running slowly. Repair Spotlight?" People make fun of the Clippy tool in Microsoft Office that tries to figure out what you're doing, and offers help when it thinks it knows your objective, but the principle is sound even if some implementations have been farcical.
Alternatively, the "Repair" item in the pop-up menu for Windows network connections doesn't get in your face without an invitation. Although you have to know enough to ask for its help (which many people prefer), it still dramatically lowers the level of knowledge and the number of user-initiated steps required to fix a problem. But all of these are just different degrees of wasting end-user time with single-user maintenance of single-user machines.
I've never needed to rebuild an index structure on eBay. I've never needed to repair a corrupted file on Google Docs. No salesforce.com user has ever needed to check for version clashes before doing a complex, possibly data-destroying upgrade of one or more pieces of a brittle software stack.
As far as the user of any competently provided service is concerned, the service "just works" – or on the rare occasions when it doesn't, it's the provider's job to figure out the problem and communicate a realistic timeline for when things will get back to normal.
We can watch, and even enjoy, the mud-wrestling between partisans of one thick-client option or another, but none of them can actually live up to the promise of "it just works." More problems? Fewer problems? That's not the point.
Making IT hiccups else's problem? That's a promise that actually can be kept, and that's what the cloud is all about.
Bookmark