Microsoft released Windows Server 2022 a few weeks ago. It did so earlier than expected, which is pretty odd, but it seems that the development of new features was already complete and mature so there was no point to wait a few months more. The last sentence pretty much explains the current status of Windows Server: so few changes, albeit interesting ones, that the product can be released months earlier than planned and without much fanfare.

While generally speaking the development of new features and functionalities for server applications mostly moved out of the base operating system and while we can also concede that Windows Server has many features at the OS level so it is difficult to design new ones, the fact each new release of Windows Server keeps bringing few new features is definitely because of Microsoft errors. Server applications today are either simple, think about a website or a Web application, or complicated. In the latter case, you generally think about microservices, Kubernetes and so on.

So we might say that since a few years Windows Server became more a tool to build base infrastructures than the base for new applications. That means that you usually install something on top of Windows Server to manage and deploy applications and you don’t usually rely on OS services only to develop your application. It is safe to say that it is a common scenario because Linux too is experiencing the same transformation.

However, Microsoft could have responded to Linux popularity by upping the ante and push more innovation into Windows Server and they didn’t.

Windows Server 2022 – continued fragmentation

The new Windows Server version has some nice and neat features. Besides a better Kubernetes compatibility, which is basically mandatory today, it features and big work to improve security. Credit where is due. That improves a lot both standalone servers security but it is also a fundamental improvement when deploying clusters, the famous infrastructures.

Other improvements include TLS 1.3 enabled by default and a few improvements around SMB, all of them improve security.

There would be other two important features to mention, one of them would be very important, but surprise: they won’t be available to all customers but only to those using Azure. At least for now.

Microsoft introduced Windows Server Datacenter: Azure Edition, another attempt to fragment Windows Server but the first one to draw a line between “standard” Windows Server and Azure versions. And we cannot welcome this kind of behavior because it basically starts to leave non-Azure Windows Server installations behind in an attempt to force Windows customers to move to Azure. Quite frankly, this is indecent especially when you factor how much Windows Server costs when compared to its competitor.

Features that Microsoft is withholding from its non-Azure customers are very important.

First, there is the much awaited hot-patching, that is a way to patch Windows without rebooting. Microsoft aims to use hot-patching especially for security patches and it would be a very ground-breaking innovation when considering how many emergency updates are needed [...]

Con un post sul blog dell’apposita sezione, Microsoft ha annunciato il supporto per Kali Linux in Windows 10, in collaborazione con gli sviluppatori del progetto. Da questo momento sarà quindi possibile effettuare il download di Kali Linux dallo Store di Windows 10 ed installarlo sul proprio PC, come parte del WSL (Windows Subsystem for Linux). Kali Linux è una distribuzione basata su Debian progettata per l’uso forense e per i test di sicurezza ed include oltre 600 tool di sicurezza e testing. Si aggiunge ad altre più note, come Ubuntu, già supportate da Windows 10. Non è stato contemporaneamente annunciato il supporto per il WSL su Windows Server, disponibile dalla versione RedStone 3.

A couple of weeks ago, Joe Belfiore confirmed that Microsoft is not “focusing” on Windows 10 Mobile anymore, essentially giving up the whole phone strategy of Microsoft. While Redmond is not officially retiring the platform, Belfiore told us that there will be no new features and no new devices, suggesting to switch to competing platforms. I almost instantly commented that it is a huge error made by Microsoft that had and has the resources to run a niche mobile platform while it waits for “the new generation of mobile devices”, as they call it.

The first casualty of that error could be the Universal Windows Platform (UWP) framework that promised to help developers to design applications for “any form factor” Windows 10 was running onto. However, with the retirement of Windows 10 Mobile and without any viable alternative for the smartphone-sized form factor, what would the word “any” mean ? While the UWP platform also includes Xboxes, IoT and holographic devices (Hololens and compatible devices), it is easy to understand that the big share of devices that could be interested in UWP apps were the desktop+mobile ones.

Xboxes, holographic and especially IoT devices for sure will feature a completely different class of apps and it will be not so common to have UWP software running the very same application on IoT devices AND desktop ones. Sure, Spotify could be useful on Xbox but since .NET is the common development framework for both Win32 and UWP there’s already a baseline of code that could be shared across the different versions of Windows-based applications and Win32 recently had access to the UWP-specific APIs too.

Xboxes, holographic and especially IoT devices for sure will feature a completely different class of apps and it will be not so common to have UWP software running the very same application on IoT devices AND desktop ones.

No-one was thinking that Win32 was going anywhere soon because Microsoft is sane and it will not cut the tie to hundreds of millions of applications just to turn itself into a new OS X but the fact that Win32 will come to ARM devices, in my opinion, makes even less important to dig into UWP ecosystem. The very fact that Microsoft not only ensured to move the Win32 ecosystem onto the boat of ARM devices but also ensured that such apps could be packaged for the store and also enabled its access to UWP APIs means that the adoption of that universal platform is not moving as fast as initially expected.

While there could be some technical reasons for that, the main point here is just that the mobile demise made less important to switch to a somewhat limited and entire new platform just to create apps that would run on desktops anyway. .NET standard 2.0 that has been recently released will help put UWP apps on par with .NET framework ones but it’s difficult to understand whom such apps are aimed to now that mobile is no more.

Don’t get me wrong: I [...]

Windows sys admins have long relied on the very wide range of tools that Windows, especially its Server versions, made available to manage the operating system. Unlike many think, such tools were not black boxes but they have been built on APIs that were available to developers, in most cases, so that they could write their own tools and access the very same information that Microsoft tools have been providing and also perform the same actions. But the tools themselves were good enough so that admins have been using them in a way or another, even when implementing their own.

Microsoft also provided Server Manager, a tool that in its latest versions allowed to manage multiple servers at once and provided access to many information and actions to perform on to local or remote servers. It also acted as a container for other administration tools when it didn’t provide the capability to perform actions on its own. At some time, Microsoft probably thought about converting its desktop tools into plugins for the newer versions of Server Manager and also started such work by integrating some of them. That was no easy task because of the wide range of utilities that Microsoft provided over time but the effort made sense.

When Core (or Nano, as of Windows Server 2016) versions of Windows Server were introduced, sys admins had to rely on Powershell to perform the same tasks since no UI was available in Core versions of the OS. Actually, some of them were quickly converting Core versions into Desktop one when they needed to perform extensive maintenance actions, reverting back to Core when done. The reason for that is oh-so-simple: performing complicated tasks using the command line is way less productive than using graphics tools, whatever Linux sys admins would like you to believe. That’s the reason why the button you must pay attention to is red, the button to do a safe action is green and you have a wheelchair icon on a reserved parking instead of printing “This parking space is reserved to people with disabilities so you are forbidden to park here unless you have a disability your own. You will receive a fine if you park here abusively.”. Metaphors. We discovered that our mind is better at metaphors long ago. Well, not all of us. We are still waiting for Linux admins to evolve but we’re confident.

Yesterday Microsoft announced Project “Honolulu”, a new way to manage Windows systems through a new shell, this time a Web-based administration shell. I had the privilege to contribute to the design and testing of the tool and I will try to describe why I loved it and why it can become very relevant for Windows sys admins.

What Honolulu is

Honolulu is a Web-based Server Manager that can be installed on both desktop and server machines and provide access, that is to say “graphical access”, to one or more servers in your environment. The first thing to notice is that Honolulu doesn’t require anything “more” [...]

Con un post sul blog ufficiale, Mark Russinovic ha ufficializzato il rilascio nei prossimi mesi (Febbraio?) del framework Coco, un framework enterprise per la creazione di blockchain. La tecnologia delle blockchain promette di rivoluzionare il modo in cui si scambiano informazioni attendibili in contesti che possono richiedere la partecipazione di soggetti esterni all’azienda o non completamente attendibili. Sempre più azienda stanno pianificando l’adozione della tecnologia blockchain per le proprie necessità interne ma spesso, secondo Russinovic, si trovano a fronteggiare dei limiti in aspetti fondamentali come le performance, la riservatezza delle informazioni, la gestione della governance e della potenza di calcolo necessaria. Questo perché le tradizionali blockchain sono pensate per l’adozione su vasta scala e con la necessità della massima trasparenza interna tra attori normalmente sconosciuti fra di loro, cosa che ad esempio obbliga a ripetere ogni transazione su ogni sistema della catena e ad usare protocolli che abbiano richieste di computazione molto intensive.

Coco è pensato per i consorzi nei quali i nodi e gli attori sono prestabiliti e controllati e questo rende possibile una alta scalabilità, massima riservatezza e una governance distribuita della rete

Queste caratteristiche, sebbene necessarie sulle reti pubbliche, non sono generalmente necessarie su quelle private o aziendali dove invece altri fattori possono essere più importanti o maggiormente desiderabili, per esempio la riservatezza delle informazioni. Microsoft ha quindi deciso di rendere disponibile il framework Coco, un sistema enterprise open-source che consente l’implementazione di reti blockchain su larga scala e con la massima riservatezza, rispondendo alle necessità delle aziende che desiderino accelerare il processo di adozione della tecnologia.

Coco è pensato per i consorzi (o gruppi) nei quali i nodi e gli attori sono prestabiliti e controllati e questo rende possibile una alta scalabilità, massima riservatezza e una governance distribuita della rete senza sacrificare le caratteristiche proprie delle blockchain come la sicurezza e l’immutabilità. In particolare Coco fornisce:

una velocità di esecuzione e una latenza paragonabili a quelle di un database; modelli di sicurezza più completi, più flessibili ed adattati alle necessità aziendali; una gestione distribuita della sicurezza e della governance della rete; il supporto per transazioni non deterministiche.

Il framework è compatibile con le blockchain esistenti ed infatti Microsoft ha iniziato ad integrare Ethereum mentre JP Morgan, Intel ed R3 hanno già annunciato la loro disponibilità ad integrare servizi enterprise basati su blockchain.

Coco funzionerà  sia in cloud che on-premise, cioè nei datacenter aziendali privati, su qualsiasi sistema operativo e hypervisor che supporti trusted execution environments (TEE)

Coco funzionerà sia in cloud che on-premise, cioè nei datacenter aziendali privati, su qualsiasi sistema operativo e hypervisor che supporti trusted execution environments (TEE) e cioè l’esecuzione di codice con il supporto di sistemi avanzati di sicurezza come Intel SGX o Microsoft Virtual Secure Mode (VSM).

La disponibilità è prevista per “l’inizio del 2018” per cui noi di TSST ipotizziamo un rilascio che dovrebbe avvenire attorno al mese di Febbraio 2018.