clearskyedge

Scale breaks even the well designed solutions. So it's very hard. At some point physics plays a factor.

I think that this will be doing with algorithms and software. This is why the edge changes the game bc the new architecture will require new thinking

In most cases, the main way to do this is to run synchronous replication between two sites that are within about 100 miles or so.

It's a business conversation. IE In transactional data, you can come to an agreement with the business that a transaction isn't confirmed until it's written offsite. That will impact the speed of transactions but guarantee data resiliency.

Agreed. Most businesses classify their applications into two categories: The applications where the cost of losing any data is so high that it's worth it to have a full sync rep implementation, and every other application where it does not

Agreed sync-rep will certainly help in this case

The technology challenge is to mitigate the impact of writing that transaction to a remote data volume.

Always-accessible data IS a reasonable expectation for an enterprise. It can't be on tape or other cold archive. You never know when you will need to recover, the data has to be at the ready. The edge can help make that happen. #clearskyedge

@MarcJKeating As your edge network grows, no-loss requires increasing attention to replication strategies to ensure continuous availability of all data, no matter where persisted. "Edge" should not be a cliff off which data is pushed to its oblivion.

@jameskobielus agreed. This is where intelligent edge caching technologies can assist. There, if the cache is destroyed no data loss ensues as the master copies are kept elsewhere safely

and are always accessible, even to other locations if necessary

the #multicloud backup and recovery is huge issue; obviously latency is key design concern

Keeping latency low and performance high. #clearskyedge

making sure the data is encrypted @ rest and in transit

Challenges in storing data at the edge including not busting the storage budget. Need the intelligence to balance fast access benefits of intelligent edge against TCO.

In many cases trying to manage data across hybrid clouds means many different interfaces and management tools needed to choreograph backup, replication, mobility, etc.

keeping infrastructure simple and scalable is critical. The apps taking up all the storage are growing like crazy, the are in multiple clouds and storage MUST be designed to support that diversity.

@DarrenMilr Yes. Ensuring consistent policy-based governance of data across myriad public and private clouds can be challenging.

@c_pallotta Right. Keeping management of diverse data across public & private clouds consistent, using simple, hopefully automated storage management tools, is a huge requirement.

Every Thursday from 1-2 ET - join us!

I'd like not to worry about this at all. It either integrates into my existing backup platform or is part of the package for a intelligent edge solution. I don't want a new solution for backup because the data is physically located at the edge.

You can SETTLE for the status quo. Primary, backup and DR are one set of challenges and can be solved together - not in silos. #clearskyedge

The edge should be used for it's purpose.. high speed data access while backups can be pushed out to the cloud for long term.

this is a great question many are learning what is best practice I haven't seen a consensus of what's a good solution

@CTOAdvisor Exactly. #managedservice #clearskyedge

A well-architected storage architecture should seamlessly replicate data from primary storage to backup/recovery edge nodes/clusters. That's are core definition of intelligent edge storage.

@CTOAdvisor Edge RPO/RTO: That's policy-driven, right? Many B/R solutions don't support edge-oriented policies well.

You CAN'T :) SETTLE for the status quo. Primary, backup and DR are one set of challenges and can be solved together - not in silos. #clearskyedge

Unless I'm using my BUs during the day for purposes other than traditional backup, I don't want them on my edge. I want them available when I need them and my edge to service my app compute.

Like all things it depends. What are you trying to protect against in backups vs. snapshots? If you have to restore from backup then do you enact your BC plan vs. recovering a corrupted file locally?

@jameskobielus replicating data from primary storage to backup/recovery is certainly one way of doing that, but it involves making, paying for, and moving copies of data

@DarrenMilr So, essentially, you're defining edge storage's primary use case as fast data access. Right?

Synthetic (do they even use that term anymore) backups from the edge to the core combined with local snapshots may meet most recovery scenarios. How often do you need to recovery TB's from backup? That's never a good situation.

Great to have you, Ellen.

Moving storage to the intelligent edge speeds data access, boosts data avaialbility, and enables more reliable disaster recovery.

Analytic applications in general. So AI/ML/IoT are but buzzwords today. Collaboration is a 2nd beneficiary.

Avoid latency issues.

@BertLatamore Right on, Bert!

Apps that monitor and triage SCADA/OT performance.

- data-intensive applications: ML, AI are perfect examples, traditional DB apps...all benefit from the intelligent edge

While we don't always consider it so, Public Cloud is also an edge location in an Hybrid-IT environment.

Also edge locations may not have strong network connections that can handle transmission of large data volumes.

@CTOAdvisor Data applications in general: intelligent edge storage can also facilitate fast in-memory OLTP at the edge.

Think @splunk, @elasticstack #clearskyedge

analytics, machine learning, or any time-critical apps that handle patient care where milliseconds can have big consquences

@CTOAdvisor Really? Can you explain how you define public cloud as an edge location?

@BertLatamore True. It's important to be in the right edge location. Highly connected colos with lots of network options.

@CTOAdvisor I agree with you Keith but utilizing both the public cloud and local DC as intelligent edge options pose data mobility issues in most cases. Then an intelligent mobility solution is in order

@BertLatamore Intelligent edge technology works best when you have excellent network connectivity and a diversity of carriers.

Bandwidth providers are still charging enormous rates for edge support. Sometimes, even if data can be moved, it is too expensive to move.

@jameskobielus If you still have a enterprise DC strategy but use Public Cloud for purpose built applications such as R&D, it's an edge location. It's massive in compute/storage capacity but it's not your central data repository.

Good one. Yes, each colocated storage facility is its own "edge" for the purpose of persisting data there for speeding access to that data by downstream users/apps.

easier to move, expensive to retrieve!

@MarcJKeating Sometimes the edge locations do not have that connectivity. A drilling platform in the middle of the Caribbean, for instance.

Bandwidth charges vary greatly based on location and the availability of carriers. For example, if you are in a well connected co-lo facility or a highly connected metro area, you'll have a far greater choice and price point for connectivity

@CTOAdvisor Oh, yes. The edge storage can refer to any purpose-built node, cluster, or zone for data persisting/access to a replicated segment of the larger central data repository.

@BertLatamore no doubt. In the case of an off-shore drilling platform, you're likely limited to satellite connectivity, which can be cost prohibitive, unless the value of moving the data in near real time was worth it to the business

It also depends on the purpose of moving or accessing data is. If you're simply trying to back up the data on the oil rig, for example, and you have a 24 hour RPO, you could send tapes on the daily supply ship.

On the other hand, if your business sees enough value in sharing that data within minutes, you could certainly obtain a high enough satellite link, but the relative cost of doing that would be much higher than the tape transport method

@MarcJKeating Microwave would be another choice, again fairly expensive. Better to analyze the data on site & send the results of the analysis rather than the raw data in many cases.