apops - Fotolia
SN blogs: Are dropped packets better than delayed packets?
This week, bloggers look into whether dropped packets are better than delayed packets, discuss cloud-based shadow IT and clarify the meaning of microservices.
Dropped packets: Is it better to drop or delay TCP packets? A podcast featuring ipSpace blogger Ivan Pepelnjak and network expert Juho Snellman discussed this very topic, particularly as it relates to using TCP versus the alternative UDP.
Pepelnjak says the answer is surprising and it may be best to focus on dropped packets.
"While you might get better results responding to increased delays (as opposed to packet drops), responding only to drops is a better survival strategy," Pepelnjak said. "[This is because] latency-sensitive algorithms back off sooner than drop-sensitive ones and thus starve in a congested network when competing with drop-sensitive algorithms," he added.
Listen to the discussion between Pepelnjak and Snellman.
Shadow IT and cloud security
Everyone knows about shadow IT in the data center. But what about shadow IT in the cloud? Enterprise Strategy Group, based in Milford, Mass., examined the impact of cloud-hosted shadow IT and its effect on operations. According to a recent ESG survey, 93% of respondents said they are aware of or concerned about shadow IT and 65% know of unsanctioned applications in use. Yet it is difficult for companies to secure apps and software they either cannot fully understand or track in their cloud networks.
To fight back, many organizations are using cloud access security brokers (CASB) to gain visibility and insight in the cloud. Nine out of 10 respondents said CASBs are critical to gain control over the use of cloud applications and thus reduce the impact of shadow IT and potential cloud breaches.
Watch ESG's video about shadow IT and cloud security.
Defining microservices
Gary Olliffe, an analyst at Gartner, sees the wave of hype and excitement around microservices continuing unabated. Because of the focus microservices have received from marketing departments, Olliffe believes that the meaning of microservices is becoming obscured, except among application platform and development tool vendors.
Olliffe said there are a number of ways to spot instances of incorrectly labeled microservices. Services implemented with small amounts of code or a simple API, services built on mutable computer infrastructure and coarse-grained or "monolithic" services built on Docker containers are all examples. Published APIs and components as well as vendor-offered modules and capabilities over which clients do not have control are other examples. "The most common misuse of the term 'microservices' is the last one: Calling a service delivered or exposed via web API a microservice," Olliffe said. "This drives me nuts because the whole point of an API is that it encapsulates the service implementation -- which might be microservices or something much less glamorous. You, the user of the API, should not know or care what architecture the provider is using," he said.
Dig deeper into Olliffe's thoughts on microservices.
Dig Deeper on Network management and monitoring
-
![]()
Construct an event-driven architecture for real-time operations
By: Mary Pratt
-
![]()
How the GraphQL Foundation is enabling data graphs
By: Sean Kerner
-
![]()
Consider these 3 fundamentals of cloud architecture
By: Michaela Goss
-
![]()
Comcast's monolith to microservices shift spotlights pains, gains
By: Fred Churchville
