Cloudy Question? There Is Gravity?

November 14, 2019

Reverse course now, bosun! After all that hype, organizations are rethinking the wisdom of remote storage. InsideBigData reports, “Companies Are Bringing Data Back from the Cloud. Now They Need a Place to Put It.” They call it “cloud data repatriation.” According to tech research firm IDC, last year 80 percent of organizations they surveyed repatriated workloads, and many expect to pull half their cloud apps to private or on-premises storage in the next two years. After racing to the cloud for several years, why the sudden U-turn? Because public clouds are not all they were puffed up to be. Guest writer and storage expert Shridar Subramanian explains:

“As well as being costly to store in the cloud, it often proves both slow and costly to download data sets from the cloud when they’re needed on-prem. The cloud also has a history of being too slow and costly for the transmission of edge data, such as unstructured data produced by the Internet of Things (IoT) devices. This unstructured data is growing at hyper speed. Indeed, IDC predicts that the total of the world’s data will increase from 33 zettabytes in 2018 to 175 zettabytes by 2025 and that 80 percent of that data will be unstructured. Those are head-spinning numbers and companies, understandably, are struggling to keep up. … All this unstructured data presents large storage and security challenges. At first, when cloud storage rose to prominence, organizations thought the answer was to move the vast majority of their data—both structured and unstructured—to the cloud. However, these same organizations soon figured out that the cloud is not only more expensive than they thought, it is also hard to access in a timely fashion when they need specific data, due to the cloud’s inherent latency.”

Rather than abandon the cloud altogether, though, organizations are going with a hybrid approach—keeping some data and apps in the cloud and some locally. And, yes, that means figuring out where to put it all. Subramanian lists some points to consider before making the move. Storage solutions should be secure, of course, and highly scalable to keep up as data grows. This means they should have the capacity to add as many drives as needed at any moment, without downtime. Ideally, they would also include analytics to help decide which information should remain close at hand and which would be better kept in the cloud.

Cynthia Murrell, November 14, 2019

How to Create Solutions in Software: The Cloud and More

November 8, 2019

DarkCyber is working on a white paper. This white paper is about Amazon AWS and its products/services for LE and intel professionals. Don’t worry, the white paper will be free to those affiliated with an enforcement organization.

In that white paper, DarkCyber’s team includes a diagram with layers. One of the reviewers of the paper told a team member:

Layers. What’s AWS? A birthday cake?

We talked about our diagram and the notion of layers. One person talked about “Layers in Software: From Data to Value.” The article included this diagram, which is different from the illustration in the DarkCyber white paper, but it conveys the same message. Here’s the Jessitron image:


The main idea is explained this way:

Feature teams need to do everything, from the old perspective. But that’s too hard for one team — so we make it easier.

This is where Developer Experience (DevEx) teams come in. (a.k.a. Developer Productivity, Platform and Tools, or inaccurately DevOps Teams.) These undergird the feature teams, making their work smoother. Self-service infrastructure, smooth setup of visibility and control for production software. Tools and expertise to help developers learn and do everything necessary to fulfill each team’s purpose. Internal services are supported by external services. Managed services like Kubernetes, databases, queueing, observability, logging: we have outsourced the deep expertise of operating these components. Meanwhile, internal service teams like DevEx have enough understanding of the details, plus enough company-specific context, to mediate between what the outside world provides and what feature teams need. This makes development smoother, and therefore faster and safer. We once layered by serving data to software. Now we layer by serving value to people.

This is a useful explanation. It applies to Amazon’s approach to the LE and intel sector. There is a twist in the Amazon digital river of products and services. That’s to be expected.

What is that twist?

The white paper will be out one the reviewers complete their inputs.

Stephen E Arnold, November 8, 2019

Microsoft Displays Its Amazon AWS Neutralizer

November 5, 2019

I read about Microsoft’s victory over the evil neighbor Amazon. What was Microsoft’s trump card, its AWS neutralizer, its technology innovation?

The answer may have appeared in “Microsoft Unveils Azure Arc, Aiming to Fend Off Google and Amazon with New Hybrid Cloud Tech.” Here’s the once closely-held diagram.


Like most AWS-hostile diagrams, it includes three features which customers like the Pentagon and other entities desire:

  1. The ability to integrate multiple clouds, on premises computers, and edge computers into one homogeneous system. (Latency? Don’t bring that up, please.)
  2. The Azure stack in one’s own computer center where it can be managed by an Azure-certified staff with the assistance of Azure-certified Microsoft partners. (Headcount implications. Don’t bring that up, please.)
  3. An Azure administrative system which provides a bird’s-eye view of the client’s Azure-centric system. (Permissions and access controls. Don’t bring that up, please.)

Microsoft has rolled out a comprehensive vision. The challenge is that Amazon and Google have similar visions.

Microsoft may want to check out Amazon’s security and access control technology. But that’s a minor point for a company which struggles to update Windows 10 without disabling user’s computers.

Great diagram though. Someone once observed, “The map is not the territory.” And then there is the increasingly relevant Argentinean writer Jorge Luis Borges who wrote:

Nothing is built on stone; All is built on sand, but we must build as if the sand were stone.

Borjes was a surrealist who could see societal trends despite his blindness.

Stephen E Arnold, November 4, 2019

Severless Framework: Good or Bad?

October 21, 2019

For any readers who have wondered whether the Serverless Framework is a good option for them, blogger Einar Egilsson provides a case study in his post, “Serverless: 15% Slower and 8x More Expensive.” Egilsson explains exactly what he tried and why Serverless was less than helpful, for his purposes at anyway. He describes his original setup, how he approached the shift (using this tutorial), how he tested the performance, and, last but not least, the pricing involved. See the write-up for all those details. The post concludes:

“I’m sure there are cases where API Gateway and Lambda are better than Elastic Beanstalk. I guess our use case is just not one of them. Maybe if you’re using API Keys, rate limiting and other stuff API Gateway provides then it makes sense to pay 3.50$ for a million requests. For us it would have been better if we could just put a normal load balancer in front of Lambda. As far as I know that’s not possible, API Gateway is necessary for http access to Lambda. But even if we were just paying for Lambda, at 10$ a day we would be paying 300$ a month instead of 164$. We have a lot of requests, but each request does very little, it’s basically one database call per request. Maybe heavier requests that use more compute time would be better served with Lambda, where you pay per 100ms of compute time. Below is a report for one request, you can see we’re using 3.50ms of compute time and being billed for 100ms, which seems like a big waste. Finally, I’m not trying to bash API Gateway, Lambda or serverless in general here, just showing that for some workloads they are a lot more expensive than boring old EC2 and Elastic Beanstalk. So that’s what we’re sticking with.”

Since the original was posted, Egilsson has amended it. Apparently, he learned a lot from the comments about what he could have done better—like using an Application Load Balancer instead of the API Gateway and upgrading to a newer instance type, for example. The software is still not right for his site, he notes, but at least he can admit, with good humor, where he went wrong.

Cynthia Murrell, October 21, 2019

Cloudy Enough Already

October 18, 2019

Cloud services have quickly become the norm, but businesses stand to lose money if they implement them without a good plan. IT ProPortal reports, “Business Are Wasting Millions on Unused Cloud Services.” Writer Sead Fadilpaši? cites a recent report from the European Insight Intelligent Technology Index that polled 1,000 European businesses. He writes:

“The report states that more than $36 million are being spent on cloud services a year. Out of that number, almost a third (30 per cent – $10.9m) gets spent on services that end up not being utilised.

When managing how they spend their money on cloud-based services, the respondents have identified three main challenges, which include determining best-fit workloads for public, private and hybrid clouds, planning and allocating budget for cloud consumption, and a lack of visibility of used services at the cst centre, workload and application level.

“‘Cloud continues to be a mission-critical enabler for agile and digital business, but it needs the right approach,’ commented Wolfgang Ebermann, President, Insight EMEA. ‘A robust operating model, that provides oversight and continual optimisation of cloud environments, is critical. Under-utilised technology has been a problem for decades, so it’s not surprising to see the problem spread to the cloud. However, by putting the right controls in place, organisations can optimise cloud consumption and ensure they only pay for services they are using.’”

The report also indicates businesses will continue to invest in new technologies, largely to put the best tech in the hands of their talented employees. That makes sense—but to avoid wasting money, companies must take the time to implement these tools efficiently.

Cynthia Murrell, October 18, 2019

Algolia: Cash Funding Hits $184 Million

October 15, 2019

Exalead was sucked into Dassault Systèmes. Then former Exaleaders abandoned ship. Algolia benefited from some Exalead experience. But unlike Exalead, Algolia embraced venture funding with cash provided by Accel, Point Nine Capital, Storm Ventures, and Y Combinator, among others.

DarkCyber noted “Algolia Finds $110M from Accel and Salesforce for Its Search-As-a-Service, Used by Slack, Twitch and 8K Others.” The write up reports that the company has “closed a Series C of $110 million, money that it plans to invest in R&D around its search technology, including doubling down on voice, and further global expansion in Europe, North America and Asia Pacific.”

The write up adds:

Having Salesforce as a strategic backer in this round is notable: the CRM giant currently does not have a native search product in its wide range of cloud-based services for enterprises, instead opting for endorsed integrations with third parties, such as Algolia competitor Coveo. The plan will be to further integrate with Salesforce although no products to speak of as of yet.

The challenge will be to go where few search and retrieval systems have gone before.

Some people have forgotten the disappointments and questionable financial tricks promising search vendors delivered to stakeholders and customers.

With venture firms looking for winners, returns of 20 percent will not deliver what the sources of the funds expect. The good old days of a 17X return may have cooled, but generating an 8X or 12X return may be a challenge.


In the course of our researching and writing the enterprise search report in 2003 to 2006 and out and our subsequent work, several “themes” or “learnings” surfaced:

  1. Good enough search is now the order of the day; that is, an organization-wide search system does not meet the needs of many operating units. Examples range from the legal department to research and development to engineering and the drawings plus data embedded in product manufacturing systems to information under security umbrellas with real time data and video content objects. Therefore, the “one solution” approach dissipates like morning fog.
  2. Utility search from outfits like Amazon are “good enough.” This means that a developer using Amazon blockchain services and workflow tools may use the search functions available from Amazon. Maybe Amazon will buy Algolia, but for the foreseeable future, search is a tag-along function, not a driver of the big money apps which Amazon is aiming toward.
  3. Search, regardless of vendor, must spend significant sums to enrich the functions of the system. Natural language processing, predictive analytics, entity extraction, and other desired functions are moving targets. Adding and tuning these capabilities becomes expensive. And it the experiences of Autonomy and Fast Search & Transfer are representative, the costs become difficult to control.

DarkCyber hopes that Algolia can adapt to these research factoids. If not, search and retrieval may be rushing toward a disconnect between revenues, sustainable profits, and investor expectations.

The wheel of fortune is spinning. Where will it stop? On a winner or a loser? This is a difficult question to answer, and one which Attivio, BA-Insight, Coveo, Elastic, IBM Watson, Lucidworks, Microsoft, Sinequa, Voyager Search, and others have been trying to answer with millions of dollars, thousands of engineering hours, and massive investments in marketing. I am not including the search vendors positioned as policeware and intelware; for example, BAE NetReveal, Diffeo, LookingGlass, Palantir Technologies, and Shadowdragon, among others.

Worth monitoring the trajectory of Algolia.

Stephen E Arnold, October 15, 2019

Cloudera Bids to Be the Next Gen Anti Financial Crime Platform

October 10, 2019

DarkCyber read “Moving Towards the Next Gen Financial Crimes Platform.” The essay, which is two parts information and three parts marketing collateral, presents a diagram of the Cloudera anti financial crime platform. The phrase “financial crime platform” could be interpreted as the airfield for dispatching a range of malware attacks, a position in which some cloud vendors find themselves either wittingly or unwittingly. In this DarkCyber article, I will refer to the Cloudera vision as an anti financial crimes platform, hopefully to make clear that the cloud vendor is not a bad actor.

In DarkCyber’s view, there are three main points about Cloudera’s enterprise focused solution. Silos of information are a problem, and Cloudera will sweep across organizational data silos, at least that’s the idea. Here are points DarkCyber noted:

  1. The focus is on the enterprise, not on a wider scope; for example, a bank, not a number of FBI field offices, each of which operates more or less autonomously
  2. Smart software (artificial intelligence, machine learning, et al) are used at the edge to provide necessary signals about activity warranting further analysis by more numerical recipes
  3. The solution can accommodate innovations either from Cloudera or from partners.

Cloudera includes a diagram of what the solution’s broad outlines are. Here’s the illustration from the Cloudera article:


Working from right to left, data are ingested by Cloudera. The content goes into an enterprise data store. A suite of financial crime “applications” operate on the data in the Enterprise Data Store and its modules. At the right hand of the diagram analytical tools (maybe like Tibco SpotFire?), business intelligence systems, and Cloudera’s Data Science Workbench allow authorized users to interact with the system.

Cloudera’s article includes this statement:

With CDP as the foundation, intelligence gaps are mitigated by a holistic enterprise view of all customer and financial crime-related data (holistic KYC), systems, models and processes.  You will also be able to tighten the loop between detecting and responding to new fraud patterns. CDP also supports open-source advances to ensure that your teams are able to experiment with and adopt the latest technologies and methods, which helps to mitigate technology and vendor lock-in.  The diagram below illustrates the Cloudera Data Platform and its various components for enterprise management. [Emphasis in the original source]

Several observations are warranted:

  1. Vendor lock is an organic consequence of putting one’s egg in one cloud-centric basket. Although it is possible to envision a system which accepts enhancements, the write and the diagram do not include a provision for this type of extension. DarkCyber posits that restrictions will apply.
  2. The diagram has “financial crime applications” without providing much “color” or detail about these policeware components. One key question is, “Will these policeware applications run “on Cloudera” or on some other system; for example, IBM cloud which delivers Analyst Notebook functions?”
  3. The write up does not provide information about restrictions on data; for example, streaming data from telephone intercept systems.
  4. Information about functional components, application programming interfaces, and programmatic methods for the platform are not provided. DarkCyber understands the need for economy in writing, but a table or a list of suggested links would be helpful.

Why is Cloudera making this play?

DarkCyber hypothesizes that Cloudera realizes Amazon’s “as is” capabilities pose a substantial threat. Cloudera wants to stake out some territory before the Bezos bulldozer rolls through the policeware market.

Stephen E Arnold, October 9, 2019

Ah, the Cloud: The Risks of Subscription Software

October 9, 2019

DarkCyber is amused when articles about the wonder of cloud-provided subscription software is presented as a real benefit to users. The team was intrigued with the information in “Adobe to Deactivate Accounts for All Venezuelan Users Due to US Sanctions.”

The write up reported:

US software giant Adobe is canceling all subscriptions and deactivating all accounts for Venezuelan users as part of its efforts to become compliant with sanctions imposed by the Trump administration over the summer.

The accounts go dead which means Photoshop, Illustrator, and other Adobe app users in Venezuela have to find alternatives.

ZDNet states: “Because of the White House’s sanctions, users aren’t eligible for refunds either.”

Developers of software which can be installed on a user’s computer may find this announcement heartening.

DarkCyber wants to point out:

  1. The cloud based service is a variation on old school time sharing. Users have limited control under certain circumstances.
  2. Government actions can have more impact when centralized services comply with mandates.
  3. Subscriptions’ benefits may not be tilted toward users.

The cloud has other attributes as well; for example, monitoring and control.

DarkCyber’s view is that the modern computing environment is becoming increasingly interesting. Those dependent on cloud based solutions may want to consider having a Plan B.

Stephen E Arnold, October 9, 2019

Amazon and Its Dark Edge

September 5, 2019

Later this year, I will make available a free essay about Amazon. Those who want the document will have to provide an email address and some other information. The essay is titled “Dark Edge.”

I am mentioning this because Information Age published “Is the Cloud and AI Becoming Two Sides of the Same Coin?” Despite the wonky subject-verb in the title, I noted a couple of points in the write up which raise issues discussed in “Dark Edge.”

First, the basic idea is that if a company embraces the cloud, smart software comes as part of the deal. That’s a good point, but it does not go far enough. In fact, most of the cloud vendors don’t think beyond generating more revenue than they did the previous quarter and figuring out how to take advantage of the dazed and confused companies trying to “reinvent” themselves. We noted this statement:

perhaps it’s time we realized that the AI and cloud computing industries are not mutually exclusive.

Yes, time. Just past time.

Second, we circled these two statements:

“Cloud adoption is motivating enterprises to undertake more proofs of concept in their firms with AI because it’s easier than ever before to get started,” said David Schatsky, managing director at Deloitte LLP.

According to Schatsky, this path is also becoming more attractive to enterprises as cloud providers continue developing AI offerings to business functions, without big upfront costs.

Easy. Cheap.

Key points.

There are a number of questions which the write up sidesteps; for example:

  1. What constitutes a win for a cloud platform? More revenue? Market share? More developers?
  2. What are the downsides for a digital environment which reflects the “winner take all” trajectory of outfits like Facebook, Google, Netflix, etc.?
  3. How quickly will one the integration of the cloud and smart software become the norm for computing, apps, and enterprise solutions?

Dark Edge will address these. Watch for the essay later this year, which assumes I will have some time to assemble our research when I am tromp around a Neanderthal disco.

Stephen E Arnold, September 5, 2019

Amazon AWS: Almost Perfect Cloud Failover Engineering

September 4, 2019

DarkCyber noted a tweet from Andy Hunt. The tweet stated:

Amazon AWS had a power failure, their backup generators failed, which killed their EBS servers, which took all of our data with it. Then it took them four days to figure this out and tell us about it.

We also noted this write up: “Strangelove Redux: US Experts Propose Having AI Control Nuclear Weapons.” Assume Amazon continues to make headway in the US government. What happens if an Amazon glitch occurs at a critical time?

Just an idle question.

Stephen E Arnold, September 4, 2019

Next Page »

  • Archives

  • Recent Posts

  • Meta