In a notable shift in its technology strategy, LinkedIn, acquired by Microsoft in 2016, has decided to abandon its 2019 "Blueshift" initiative, which aimed to transition its data center operations from on-premises infrastructure to Microsoft's Azure cloud services. This reversal indicates that LinkedIn will continue to rely on its existing on-premises data centers for the foreseeable future.

The decision represents a PR challenge for Microsoft, which has been advocating for the migration of on-premises companies to its Azure cloud platform as a key component of its cloud strategy. However, this setback may be rooted in the unique complexities of LinkedIn's cloud infrastructure. Reports suggest that LinkedIn's distinct system architecture would require extensive code modifications for a transition to a public cloud environment like Azure.

This complexity underscores the challenges some companies face in adapting legacy systems to modern cloud solutions. The $1 trillion question is which legacy systems will resist full cloud integration due to technical and operational hurdles.

- Amazon Q - An AI assistant for AWS environments, costs $20/mo. Takeaway: A lot depends on the execution but the idea makes sense - an all-knowing assistant to handle your ever-evolving cloud environment.

- Salesforce Partnership - Salesforce will integrate many of its products with AWS, and start offering its services on AWS Marketplace. Takeaway: Microsoft has gained a lot of ground by bundling its Azure cloud offering with Office, Teams, Dynamics, and other products. This AWS-Salesforce partnership may offset those gains a bit, though it comes late in the game.

- Graviton4 - New ARM server hardware with better performance - Amazon says it's up to 40% faster for databases, 30% faster for web applications. Takeaway: AWS continues to heavily invest in its own silicon, allowing them to reduce cost in some cases, and achieve a better negotiating position vs vendors in others.

- Trainium2 - New AI training hardware. Takeaway: Trainium2 was designed before the ChatGPT explosion and AWS is still behind the competition but the company is investing heavily in AI hardware. As always, Amazon is hedging its bets by also partnering with Nvidia and purchasing a large group of H200 GPUs.

In a recent announcement, Microsoft revealed plans to phase out MariaDB from its Azure Database platform. Starting from the latter part of 2023, it will no longer be possible to create new MariaDB databases on Azure. By September 2025, Microsoft intends to have completely discontinued MariaDB support. The company is guiding its customers towards Azure Database for MySQL – Flexible Server as an alternative, highlighting its superior features, enhanced performance, and more efficient cost management.

Originating as a fork from MySQL in 2009, MariaDB has maintained compatibility with its predecessor. However, in the realm of open-source relational databases, it now ranks behind MySQL and PostgreSQL — the latter steadily climbing to become the market's leading choice.

The move to mandate cloud migrations for MariaDB users on Azure is a relatively uncommon occurrence, suggesting that MariaDB's adoption on the platform may have been modest.

Amazon Web Services (AWS) has announced plans to launch an EU sovereign cloud platform. This platform is specifically designed to allow European customers to store data exclusively within the region. Although AWS already operates multiple data centers in the EU, this existing infrastructure is interconnected with other regions and might rely on tools set up in non-EU countries.
The new independent cloud infrastructure, located in Germany and managed solely by EU personnel, will resemble standalone AWS deployments like AWS Outposts. This design enables organizations to utilize the infrastructure without concerns regarding the geographical location of their data.
The decision to locate the new cloud region in Germany is strategic, as the country has been more adamant than most about data locality and preventing US-based companies from accessing its sensitive data.

Attackers continuously refine their methods, launching increasingly aggressive denial of service attacks against cloud providers and major websites, often with the aim of extortion or intimidation. Recently, Google, Cloudflare, and AWS disclosed that they faced a novel distributed denial of service (DDoS) assault utilizing HTTP/2 stream reset mechanisms.

In essence, the attacker establishes an HTTP/2 connection that can encompass multiple streams. Upon opening a stream, the server commits resources to manage it. If the stream issues an RST_STREAM (stream reset) command, the server must relinquish those resources and provision new ones. By continually initiating and then resetting these streams, attackers can overburden the server using the same number of attacking devices.

In late August, peak attack rates surged to 398 million requests per second for Google, 201 million for Cloudflare, and 155 million for AWS. Due to the swift and expert response from the affected providers, users experienced no disruptions.

Choice Hotels, an international hoteling giant, will close its last on-premises data center later this year, completing a multi-year transition to the public cloud. The hospitality company decided to adopt AWS over a decade ago, first moving its reservation system in 2014. Large-scale infrastructure migration started in 2019 and is set to complete within a few months. 

Choice took a "lift-and-shift" approach, modernizing some systems while simply porting others. The prolonged transition reflects the inertia of legacy infrastructure. While many advanced workloads have moved to the cloud, plenty of compute remains on-premises.

Traditionally, Oracle's database services have been tethered exclusively to its own cloud platform, OCI, which holds a modest market share. However, in a decisive response to the ever-evolving cloud market, Oracle has entered into a strategic alliance with Microsoft, launching Oracle Database@Azure. This partnership means Oracle's database services will now be accessible within Microsoft Azure data centers, a pioneering step that integrates Oracle's hardware directly within Microsoft's infrastructure for the first time.

Amidst a discernible decline in Q1 2023 revenue and with rivals like Snowflake and Databricks consolidating their positions across all primary cloud platforms, Oracle's collaboration with Microsoft signifies a robust strategy to reclaim lost territory in the database and data warehouse domains. Despite this new development, AWS and GCP are not expected to offer similar services like Redshift or BigQuery beyond their own ecosystems.

Microsoft's Azure cloud services faced significant downtime on August 30th, affecting numerous services including Azure, Microsoft 365, and Power Platform for over 24 hours. The root cause? A combination of a utility power sag in Australia’s East region and a subsequent failure of cooling units within a data center.
Ironically, while the technology faltered, the human factor magnified the issue. With the cooling units offline, manual intervention was needed to prevent further damage. However, there were only three staff members present that night at the data center campus, making it impossible to timely restart the failed cooling systems. In a candid admission, Microsoft acknowledged the lack of personnel, stating, “Due to the size of the data center campus, the staffing of the team at night was insufficient to restart the chillers in a timely manner."
In the aftermath, Microsoft has temporarily bolstered its night staff from three to seven members. Alongside human resources, the tech giant also promises improvements to its automation systems to counteract similar incidents in the future. This outage joins a series of service disruptions Microsoft has reported in recent months, emphasizing the critical role of both technology and human oversight in maintaining operational efficiency.