Welcome to The Long View—where we peruse the news of the week and strip it to the essentials. Let’s work out what really matters.
This week: Intel optimizes Linux multithreaded networking, data centers in space (again), and more DevOps toolmaker layoffs. (more…)
Amazon Web Services (AWS) has expanded the number of virtual machines based on the latest generation of its Graviton processors to include instances that support DDR-5 memory.
The latest Amazon EC2 instance types, dubbed M7g and R7g, that are now generally available are based on the AWS Graviton3 processors developed using an architecture created by Arm. Previously, these instances were only available as part of a technology preview program.
Rahul Kulkarni, director of product management at AWS, said these latest offerings take advantage of DDR-5 memory to provide 25% better performance than the previous sixth generation of Graviton instances. DDR5 memory increases the amount of bandwidth available by 50%, he noted.
Graviton processors now provide the lowest-priced instance of virtual machines in addition to the best performance and overall price/performance across the AWS portfolio, Kulkarni added.
AWS is making a case for employing its AWS Graviton processors for general-purpose workloads as more IT teams are being tasked with reducing costs as the overall number of applications being deployed in the cloud continues to increase. In fact, all of the top Fortune 50 companies have already deployed workloads on Graviton processors, noted Kulkarni.
While it’s not clear how many applications are being developed to run natively on Arm processors, the process of refactoring existing applications to run on Arm processors is not as intensive for most applications as it was when previously moving applications from one class of processors to another.
In addition, Kulkarni noted AWS has committed to providing a full range of DevOps services, including a continuous integration/continuous delivery (CI/CD) platform, to help facilitate any transition to Graviton processors.
There are, of course, multiple strategies for containing cloud costs that range from committing to consuming a specified amount of compute resources annually at discounted rates to relying more on spot instances that are available for limited amounts of time. Regardless of approach, the days when developers were allowed to invoke cloud resources at will appear to have come to an end as enterprise IT organizations become more aware of the total cost of IT. As a result, many of them are adopting financial operations (FinOps) best practices to maximize the utilization of cloud infrastructure resources.
AWS, like most cloud service providers, gives organizations a wide range of processor options. However, it’s apparent that AWS views Arm processors in the form of Graviton to be the most efficient from both a price and performance perspective. It’s not clear how quickly organizations are embracing Graviton in the cloud given the dominance of x86-based platforms, but it’s apparent there is now a clear alternative for general-purpose workloads. If fact, the more data-intensive those workloads become, the greater the interest in more efficient processor architectures.
Of course, x86 platforms are not going away anytime soon. But as more Arm-based processors find their way into the enterprise, DevOps workflows will only become more challenging to manage.
Data democratization has become critical for organizations to leverage data’s true value and fully realize its benefits. Making data available across the organization helps companies to serve customers better and make data-driven decisions that align with their goals and objectives. It essentially means enabling access to a company’s data resources to employees, bound by reasonable limitations on legal confidentiality and security.
As a means to remove data silos in organizations, the indispensability of data democratization is clear. It allows more widespread access to and use of data within an organization. This can lead to more informed decision-making, increased collaboration and innovation and more efficient use of resources.
Breaking down data silos can help reduce the risk of data breaches and improve data security. While the jury may be out on whether data democratization is beneficial in preventing security breaches, silos between networks or security systems make it difficult to identify large-scale attacks as they tend to represent disjointed security protocols. There is less likelihood of robust security protocols between non-communicating teams.
New-age technology such as cloud computing, blockchain, and advanced analytics are aiding data democratization. For example, 5G enables more significant volumes of data to be moved faster and more securely in the form of big data. Blockchain provides higher-quality data and improves the operation of cyber and physical systems. In addition, IoT allows for exchanging data between devices, where processes can be established between devices without intermediaries.
A hybrid cloud adoption survey conducted in March 2022 of IT professionals primarily based in North America and Europe revealed 93% of the IT industry would adopt a hybrid of cloud and on-premises solutions or migrate fully to the cloud within five years, using it to store and manage data. Gartner estimated that 85% of enterprises would embrace a cloud-first principle by 2025 and that it will be essential to execute digital strategies in business.
This suggests that new-age technologies play a significant role in breaking down data silos and enabling organization-wide access to data. By leveraging these technologies, organizations can easily store, process and share data, leading to better decision-making, increased collaboration and improved business outcomes.
Data creation, development, and utilization directly relate to Industry 4.0 applications. In practice, data democratization is seamlessly transferable to Industry 4.0 applications. The question arises: Do industrial companies have the capabilities today to apply the concept of data democratization? And to what extent?
As new-age tech drives the emergence of new business models, companies that will lead a cultural shift as well as a technology evolution will be successful. This is possible via a business-centric data strategy, along with the democratization of analytical tools and platforms to enable stakeholders to help build better insights into their business processes.
A Google Cloud and Harvard Business Review survey of industry leaders revealed 97% believe organization-wide access to data and analytics is critical to the success of their business. However, only 60% of respondents believed that their organizations effectively distribute that access today.
What are the prerequisites for data democracy? First, companies should inculcate a culture change to create a pull from inside the organization rather than first developing the technology. Second, decision-making in companies is based on experience and intuition than on data; they should compulsorily begin to use data as a baseline for business decisions. Third, leaders should encourage a data-driven culture to grow from within.
Formalizing and sharing knowledge can remove roadblocks to the democratization of data. Therefore, one form of action is to set specific incentives to foster a mindset of knowledge sharing. For instance, existing practices to incentivize participation in the suggestion or lessons-learned programs can be used as a reference.
Many companies have procedures to involve employees in the continuous improvement process. But while employees can use the existing IT systems to perform their day-to-day work, they need more capabilities to cover their individual information needs. Thus, as a way forward, capability building within is critical. Another strategy is decentralizing decision-making; if employees can gain insights independently, they should be able to act on these insights independently.
Organizational culture is vital to data democratization, as it determines the attitudes, behaviors and values that guide how data is shared and used. On the other hand, a culture that values secrecy or hierarchical control over data can lead to resistance to data democratization efforts and perpetuate data silos. A culture that lacks trust and transparency can also lead to reluctance to share data, which can impede collaboration and hinder data-driven decision-making.
Organizations need to assess and work on their tech stack and internal culture to promote a data-friendly culture supporting data democratization. To foster a data-friendly culture, organizations should encourage open communication and collaboration among employees and provide training and resources for employees to access, understand, and analyze data.
Open communication, collaboration, innovation and trust are essential attributes for breaking data silos. A corporate culture that values data-driven decision-making will lead to more widespread access to and use of data and support the implementation of new-age technologies. Finally, there is a need to establish clear data governance policies and procedures that support data democratization and build a continuous improvement process by encouraging experimentation and learning from failures.
Sysdig this week announced there is now a Wireshark Foundation that will oversee future development of the widely used open source traffic protocol analyzer.
Gerald Combs, who is now director of open source projects at Sysdig, along with Loris Degioanni, CTO of Sysdig, launched Wireshark in 2006 to enable teams to monitor network traffic, learn protocols and packet basics and troubleshoot network problems. There are more than 2,000 contributors and Wireshark has been downloaded more than 60 million times in the last five years alone.
The Wireshark Foundation will provide a framework for long-term stewardship and sustainability. The foundation will be led by Sheri Najafi as executive director, and Janice Spampinato, Hansang Bae, Najafi, Combs and Degioanni will serve on the board.
Combs said rather than donating Wireshark to an existing foundation, it was decided to set one up to formalize many of the community processes that have evolved over the past two decades.
Degioanni said the foundation will also ensure Wireshark remains relevant long after its creators eventually retire. Wireshark has a graphical user interface (GUI) that provides IT teams with an easy-to-understand traffic protocol analyzer. However, as applications become more distributed, additional types of IT professionals, including DevOps teams, are analyzing network traffic. Similarly, cybersecurity teams are using these tools to help track down the source of anomalous activity.
It’s not clear whether network operations (NetOps) and DevOps workflows are merging, but the level of collaboration has increased. As networking increasingly becomes software-defined, it’s clear that developers will employ application programming interfaces (APIs) to manage networks as code in much the same way they do infrastructure in the public cloud. Most of those networking APIs are comparatively low-level, but with the rise of services meshes, it’s clear networking services are becoming more accessible.
The biggest challenge is distinct cultures that have arisen around networking and application development. IT organizations will need to decide whether to provide development teams with the ability to programmatically invoke networking services on their own or through a self-service portal managed by a NetOps team. Regardless of who is responsible for managing the network going forward, the divide between DevOps and NetOps is already starting to narrow. The biggest challenge is simply finding a common language to express core concepts that eliminates any sense of fiefdom that might exist between them. After all, any conflict that arises can usually be attributed to a simple failure to communicate.
In the meantime, in the age of edge computing, network latency issues are becoming even more challenging. An application based on microservices architecture is especially sensitive to network latency issues that are difficult to troubleshoot over an extended network. Tools that make those issues more obvious are crucial.
Welcome to The Long View—where we peruse the news of the week and strip it to the essentials. Let’s work out what really matters.
This week: Don’t get scammed looking for a job, and momentum grows for the 32-hour working week. (more…)
Welcome to The Long View—where we peruse the news of the week and strip it to the essentials. Let’s work out what really matters.
This week: Working from home is here to stay, and Agile is still a failure. (more…)
Tabnine today announced it extended its generative artificial intelligence (AI) platform that enables developers to write code faster to now include the ability to automatically generate tests for that code.
Available in beta, this capability promises to make it easier for developers to test code at the time it is being developed. That approach should make it simpler for developers to address issues without losing any context.
Brandon Jung, vice president of ecosystems and business development at Tabnine, said this latest addition to the Tabnine platform is part of a larger effort to extend the reach of the Tabnine generative AI platform across the entire software development life cycle. The testing capability is expected to be generally available in the third quarter, and Tabnine expects to also employ generative AI to automate a much wider range of tasks across the software development life cycle within the next 12 months.
The Tabnine platform supports multiple programming languages including Python, Java and JavaScript and is designed to be integrated with integrated development environments (IDEs) such as Visual Studio Code and Jetbrains. The company has also previously integrated its code completion tool with the GitLab continuous integration/continuous delivery (CI/CD) platform. The overall goal is to make it easier for developers to automatically write code based on custom models using approved source code hosted in a secure private repository, said Jung.
Generative AI creates a large language model that assesses the probability of what the next line of code will be based on what has preceded it. Tabnine is now extending the model it created to also generate tests for the code its platform helped a developer write. It’s not clear what impact this ability to automate test creation will have on the need for dedicated testing teams, but it is about to become considerably easier to shift more responsibility for testing further left toward developers.
It’s not likely DevOps teams will be replaced as generative AI is extended across workflows, but the overall size of those teams might shrink, noted Jung. DevOps teams will soon be able to do far more with fewer people, he noted. At the same time, the barrier to DevOps adoption will also fall as AI platforms make it simpler for more organizations to embrace DevOps best practices, added Jung.
The challenge will be making sure the data collected to train AI models is of a high enough quality to ensure the desired outcome, noted Jung.
One way or another, it’s now only a matter of time before generative AI capabilities are applied more broadly. Platforms such as ChatGPT developed by OpenAI are only the tip of the iceberg that impacts almost every manual process, including software development and deployment. The issue will be determining at what rate those innovations will become practical to employ. In the meantime, it is already apparent generative AI platforms are having a significant impact on the rate at which code can be developed. Inevitably, that means the amount of code moving through DevOps pipelines at any one time is about to increase significantly.
Logz.io has integrated the ChatGPT generative artificial intelligence (AI) platform with its Open 360 observability platform to extend its automated recommendation capabilities to reduce mean-time-to-remediation.
ChatGPT, developed by OpenAI, is based on a large language model that Logz.io uses to enrich crowdsourced data. That approach enables ChatGPT to surface links to related information and best practices for resolving IT issues.
Logz.io CTO Asaf Yigal said the company has integrated ChatGPT with its existing AI engine, dubbed Cognitive Insights, that applies machine learning algorithms to identify anomalies indicative of potentially critical IT issues. The company expects to be able to add other generative AI capabilities as additional advances are made, he added.
The Logz.io Open 360 platform is based on open source OpenSearch, Prometheus and Jaeger software that the company has infused with machine learning algorithms. The integration with ChatGPT is enabled via an application programming interface (API) that OpenAI has provided.
The overall goal is to employ AI to surface the most relevant issues a DevOps team needs to focus on without having to search huge volume of logs, tracing and metrics data, said Yigal. Generative AI platforms solve a fundamental cognitive load problem created today as DevOps teams collect more data, noted Yial. The more data a DevOps team collects, the harder it becomes to determine what’s going on in an IT environment, said Yigal. He added that generative AI platforms make it possible to analyze massive amounts of data at a scale that isn’t otherwise achievable.
AI tools should make it possible to manage IT at a level of scale that eliminates many of the low-level data engineering and analytics tasks that previously required manual effort from a DevOps engineering team. It’s not clear how many other DevOps tools and platforms will be invoking generative AI capabilities in the months ahead, but DevOps teams should assume that large language models will soon be augmenting DevOps teams in ways that previously would have been thought unlikely.
For example, discovering vulnerabilities in software running in production environments is about to become trivial as artificial intelligence (AI) platforms such as ChatGPT evolve. A generative AI platform can identify code vulnerabilities in much the same way as a traditional scanning tool. An AI platform, however, could theoretically monitor code repositories and scan them for vulnerabilities as updates and commits are being made in real-time. DevOps teams should be able to use generative AI platforms to discover and remediate these issues before code is deployed in a production environment.
It’s just as probable bad actors will use the same platform to achieve the same goal, so from that perspective, many DevOps teams will now find themselves in an AI arms race to improve DevSecOps workflows.
There are still lots of issues to be worked out in terms of the overall accuracy of generative AI platforms, but they will only become more intelligent as processors continue to advance. The issue now will be determining precisely where the man-machine interface for DevOps teams is today and in the future.
Software life cycle management is becoming more critical for today’s technology-fueled organizations—because you can’t just release software into the world and expect it to work perfectly for years to come. More often than not, development teams must constantly adapt their software, progressively add features and cater to new use cases as they arise. This is nothing new for web APIs, which are prone to continual evolution and whose integrations require regular upkeep to avoid breaking changes.
As such, the discipline of API life cycle management has gathered steam. This concept focuses not only on the design and development of APIs, but on the entire life cycle, including testing, monitoring, documentation, maintenance, versioning, CI/CD and even deprecation. Localizing a strategy for API life cycle management from the get-go can increase the longevity and usability of a platform. It also helps software providers treat their APIs as an iterable product rather than a one-and-done offshoot of a digital platform.
I recently met with Rakshith Rao, co-founder and CEO of APIwiz, to gather insights on standardizing API management across an organization. According to Rao, there are many ways to implement governance to make these interfaces more consistent. For example, automated security analysis, contract-driven development and implementing style guides are just a few ways to wrangle an increasingly extensive API portfolio. Below, we’ll review these tips and others to consider how to best manage API life cycles, hopefully, before sprawl sets in.
Medium-to-large organizations might be handling thousands of APIs, said Rao. This new paradigm brings many opportunities for digital transformation, but it can also be a nightmare to manage. Too often, APIs aren’t built with tangible business requirements taking center stage.
Many of these initiatives didn’t begin with consistent policies from the get-go. For example, it’s also easy to skip out on documenting internal services. But, fractured practices harm discovery and reusability. To rectify this issue, Rao shared many ways in which technical leads can promote better API hygiene.
The API life cycle spans stages of design, deployment, deprecation and eventual retirement. And throughout this life cycle, developers will likely release new versions of the same service. But although you want consumers to adopt the latest version, some large organizations must support legacy versions for years to come, explained Rao. As a result, different API versions are often developed and run simultaneously by various internal or partner teams. Sometimes, “no one turned the tap off and the water’s still running,” he said.
To enhance visibility for all active (and non-active) sessions, it’s important to conduct ongoing monitoring. This will help discover individual client usage patterns and help avoid shadow or zombie APIs from forming. These forgotten endpoints could pose security vulnerabilities, said Rao, as you might not enforce best practices on older versions if you don’t know they exist.
Not having consistency between APIs could pose data compliance issues. One method to avoid errors is to conduct design reviews on an ongoing basis, and to automate this security analysis at the design level, said Rao. Perhaps authentication tokens need to be encoded, logging requests aren’t coming in, or payloads are structured incorrectly. Whatever the case, having a complete checklist to scan and performing ongoing security analysis early on to catch these inconsistencies can be a big boon to overall governance and long-term cybersecurity.
Very simple data modeling errors could severely impact a system, said Rao. Thus, it’s in an organization’s best interest to use standard taxonomies and ensure data governance is an overall aspect of life cycle management. One way to ensure consistency is by using specification-driven development—OpenAPI is a common standard for this need.
Rao also suggested developing API design style guidelines but admitted that these guides become challenging to enforce and limit as an organization scales. This is because various teams in different departments will have different API needs and often design their interfaces for various front-end requirements. His takeaway is to centralize practices but take a GUI approach to configuring and sharing them. This way, API style can be “minimally intrusive but uniformly enforced.”
A full 98% percent of enterprises are using at least two cloud infrastructure providers, according to a recent Oracle report. Nearly everyone is adopting a multi-cloud strategy, but this new reality brings novel pressures for DevOps, especially when overseeing the deployment of APIs across different cloud platforms. An API life cycle management platform decoupled from these CSPs could help wrangle polyglot deployments.
API life cycle management goes well beyond the development and runtime phase. And as more API-first development occurs within organizations, maintaining standard interfaces will naturally require more effort. Rao also advocates for increased infrastructure automation and low-code solutions to meet these new challenges. Automating various parts of the API life cycle so there is less manual work “helps to bring down the guesswork of ensuring reliability when you release an API,” said Rao.
We’re seeing some enterprises move from supporting hundreds to thousands of APIs. In this new paradigm, “API sprawl is imminent,” said Rao. As such, it will become virtually impossible to manually keep these interfaces in line, understating the need for more automation. Especially during times of economic uncertainty and leanness in the tech market, the democratization of such API management automation will be vital in helping organizations move faster.
Having broken pieces of information and scattered API contracts will lead to a messy and potentially insecure adoption. As such, API management can no longer be done using sticky notes on the edge of a computer screen, said Rao. Taking the guesswork out, centralizing common policies and shifting more legwork behind the scenes will be required to avoid a splintered portfolio of services.
Welcome to The Long View—where we peruse the news of the week and strip it to the essentials. Let’s work out what really matters.
This week: Denis Pushkarev is fed up with core-js freeloaders, and hundreds more malicious packages found at PyPI. (more…)