How to Set Realistic Recovery Point Objective (RPO) Goals

You found the backup wrap up your go-to podcast for all things

backup recovery and cyber recovery.

In this episode, we're talking about recovery point objective or RPO, which

is how much data that you're willing to lose when things go sideways.

Spoiler alert, most people's RPOs are complete fantasy.

I mean, you think you can only lose an hour of data, but

you're backing up once a day.

That's a problem.

We'll break down what RPO really means, why it's measured in time and not, uh,

the amount of data and how ransomware can totally mess up your carefully planned.

Objectives.

Plus I'll share some practical ways to rightsize your RPO.

Talk about database transaction logs and explain why your SaaS apps need the

same love as your on-premises systems.

Let's talk RPO.

By the way, if you don't know who I am, I'm w Curtis Preston, AKA, Mr.

Backup, and I've been passionate about backup and recovery for over 30 years.

Ever since I had to tell my boss there were no backups of

that database that we just lost.

I don't want that to happen to you, and that's why I do this.

On this podcast, we turn unappreciated admins into Cyber Recovery Heroes.

This is the backup wrap up.

Hi, and welcome to the backup wrap up.

I am your host, w Curtis Preston, AKA, Mr. Backup, and I have with me a guy who I

called while laying upside down yesterday.

Prasanna Malaiyandi, how's it going?

Prasanna,

I am good Curtis, and do you wanna tell the listeners what you were doing?

Laying upside

I was underneath my Tesla for the first time since I bar, uh, borrowed

it since I bought, bought it.

Two years ago, uh, I, I did a, I made a boo boo and I, um, I did a, you know, over

here in California, I think it's worse here in California than other places.

We have these, like, you know, when you go through an intersection, there's the big

dips before and after the intersection, and you, you seem to, I, I just think.

We don't get rain, but we get it.

We get it in torrents.

And so they have these like huge dips at the beginning of many intersections.

And if you're not paying attention, you can easily bottem out.

And I apparently, I found out after the fact that I apparently bottemed

out so hard that the two bolts that held the little, um, they're

little 10 mil, 10 millimeter bolts.

But the two bolts that hold on this.

Like what?

What's that?

It.

Under tray.

So under trade, it's like protects the undercarriage of the car.

Uh, I just sheared them off.

And then for apparently a while, it had been held on by two other 10

millimeter bolts, which weren't screwed into anything other than the plastic.

Uh, fascia, right?

And so then at some point that, uh, you know, didn't work.

And then, uh, so that came off.

So anyway, so I had to rip all that off and put it all in.

And it wasn't until I did all this, I, I bought a new shield and

I went to go screw it in there.

And by the way, that meant lifting up a Tesla, which for the record.

Ain't no walk in the park.

Yeah.

that was interesting.

But then I called you, then I called you.

I was like, FaceTime.

And you're like, uh, because I don't normally FaceTime you.

You're like, what am I looking at?

I'm like, uh, look at the underside of my car.

It is really weird though, right?

Like looking inside, something like that, that like, like I kind

of know what I'm looking at when I'm looking at a gas car, right?

But there's all these parts and none of them.

Are familiar, right?

I'm like, okay, I, I understand steering parts, right?

Um, and, um, uh, and so I had to, and, and did, I did, ultimately I had to buy.

What, what, what's the, what's the part called the, uh.

bar.

The, well, the, the stabilizer bar, but then the bushing for the stabilizer bar

and then these two brackets that hold the bushing that hold the stabilizer bar.

That's what I had to buy.

And uh, amazingly I got them for $15 each from Amazon shipped and two days for free.

Ugh,

Nice.

nice.

Um.

that is gonna be your job.

That is gonna be my job, luckily.

Yeah.

Now, now that I've figured all that stuff out, it, it'd probably be 20 minutes.

Um, the hardest part will be lifting up the car.

Make sure that you have all the hardware before you take stuff apart.

I, I, yeah.

Okay.

Yeah.

What do you think you're dealing with here, Mr.

just make sure that you have enough, 'cause I don't know what

you actually got with the kit versus like what you might need when you

All I, all I need is the bracket.

I got all the other stuff right.

The thing is that when I went to do this five minute job,

I realized that I had to take, I had to do.

Like I had to take off other parts to get to because I realized basically

as I did this more and more, and I realized that by the end, by the time

I was in my, my recovery point, uh, was very different than, uh, what I

originally had, had, had envisioned.

But, um, because, uh, just basically I lost so much more than I had originally

planned to lose underneath my car.

Yeah.

So currently my car looks worse underneath than it did when I started.

Um.

But that's okay.

Sometimes you gotta take a couple steps backwards in order to move forward

Sure we'll do that.

something like that.

Hopefully you don't do that with a, with a, with a recovery point.

So today we are talking about recovery point objective, which I would

define very quickly as saying it.

It is just how much data we agree we're allowed to lose as measured by time.

Right.

So it's not like what.

Why would you ever lose data?

Curtis isn't backup, supposed to be?

Never lose data.

Yes.

Yes.

We'll get to that.

Uh, and also why, again, just like RTO, most people's RPOs

are complete fantasy, right?

Um, so.

Uh, basically again, it's as measured by time, so it's not like we agree

we're gonna lose, um, 10 gigabytes of data or 10 terabytes of data.

We agree that we're gonna lose or allow to lose up to 12 hours of data, 36

hours of data, whatever the number is.

And again, just like with our TO it, uh, which we just did an

episode on recovery time objective.

If you didn't see that one, then go, you know, uh, and again, you can watch

these either on YouTube or you can listen to 'em on your favorite pod catcher.

Um.

Is that different scenarios, different recovery scenarios.

We'll probably call for different RPOs.

Right.

Uh, a ransomware scenario is probably, once again, possibly you're going to

have to accept more data loss than you would in just a regular recovery.

Right.

Regular loss of a server or whatever.

Right.

Well, because you might.

Find out that even your, some of your backups are corrupted, right?

That, that you've been backing it up for a week and it was, some part of

it was encrypted two weeks ago, right?

And so you might, you might have to recover to some, to some older ba you

know, from some older backup, right?

Uh, again, that, that is something that, that can happen, uh, as a, as a recovery.

Um, and

I,

go ahead.

I have another question on RPO.

Yeah.

So it's the recovery point.

You said it's the amount of data you agree to lose, right?

Is it the amount of data you agree to lose since your last successful backup?

Or is it the amount of data that you're willing, like, can

you define that a

bit clearer?

great, great great question.

So it's the amount of data that we, uh, agree to lose.

Period.

Right.

Uh, as measured by time, what will determine the amount of data you

actually lose is the last successful backup that you're able to use.

That actually rhymed.

Let me wrap that.

Um, so.

Your, your, your RPO happens, let's say, um, basically it's

the, the time is measured.

It's measured backwards.

Uh, whereas our RTO is measured forwards from the, uh, outage.

RPO is measured backwards.

So from the point of the outage.

How far back are we allowed to go and still consider it to be successful?

So if I, if this is a database and we're using, uh, redo logs and transaction

logs, hopefully you can actually restore right up to the point of failure, like

right up to just before the point of failure, even in a ransomware scenario.

Right.

Yeah.

Uh, because.

Generally with databases, if you start encrypting it, it's

gonna, it's encrypt everything.

Or you know, the moment you encrypt any part of the database, the whole

database is gonna crash, right?

So if you've got transaction logs and those transaction logs are being

protected, key thing there, right?

And they're being shipped off to some other system that hasn't been

attacked, then, um, which would be part of your recovery system, right?

Yep.

Then you should be able to just go back minutes, right?

Uh, again.

That's just the restore, right?

It's gonna take a while to figure out which, which things we're gonna

restore if this is a ransomware event.

But if it's a file on the opposite end of that, if it's a file system,

Mm-hmm.

you may have sys, you may have files in there that have been getting

encrypted over time for months.

Right?

Um, because the dwell time.

Do you want to, you wanna define dwell time?

Yeah.

The dwell time is how long ransomware sits in your system before it starts

doing something or before it's detected.

Yeah, so the dwell time may be measured in months.

There was, we, we covered one, uh, a little while ago that it was like a year.

Right.

Um, and.

A.

Yeah.

Yeah.

That, yeah, that was literal, that wasn't that long ago, but

I was like, which one was that?

That was like three weeks ago that we did that.

Um, yeah, that was an interesting story, right?

Where, where it happened over a year.

And so if they, if they're just, if they're really trying to mess

with you, they're going to encrypt little files here and there.

Possibly ones with older, um, access times, right?

That haven't been looked at in a while.

Um.

And so that's gonna be very complicated, right?

And you're, now that I think about it.

The RPO is almost an irrelevant concept there because we typically talk about RPO

from a server standpoint or an application standpoint, or a file system standpoint,

but for a file system that has been being encrypted over time, the RPO is actually

going to be many, many little RPOs.

Right, right,

because you're always looking for what's the valid data and

try to pull the newest data

right.

the

Which may reso, which may actually be thousands, potentially tens of

thousands of individual restorers rather than, uh, and hopefully you

can script that, uh, in the, in the book, um, that we, that we are, we are

finishing the editing of right now.

Literally.

Yeah.

Uh, that would be learning ransomware response and recovery.

I actually wrote a little script that could, that could basically

comb your way through a file system.

It's a very basic script, but it's just an idea that it could give you, if you

could comb through the file system, find the files that are encrypted, and

then find the oldest or the find the most recent version of that file that

wasn't encrypted and restore that file.

So you're, you're actually doing many little restores and

hopefully you can automate that.

Yeah.

But the, the, the point is that your, your RPO is the, that amount

of time that you agree that you can, you know, uh, how much you can lose.

Right?

The RPO is measured.

Going backwards in time from the incident , we can say we're gonna lose three hours

worth of data, whatever it is, one hour's worth of data, two weeks worth of data,

whatever time you've agreed on that is what your recovery point objective is.

Whether or not you can meet that or not would we would call that

recovery point actual, right?

Um, and so the difference would be, you know, again.

The, yeah, the gap between the two,

Yeah.

potentially an issue, which you might need to look at now.

One thing I wanted to ask you, Curtis, is like as a backup, if I was a backup admin,

Yeah.

I, I do not play a backup admin either on TV or on this podcast or

anywhere else, just to be clear, right?

But as a backup admin, am I the one just sort of going to be like, Hey, yeah, I

think we can lose like one hour of data.

Like

Y

who is making that decision?

Yeah, great question.

Just like RTO, the answer is absolutely not right?

You should never be making any procedural decisions like that, right?

Um, this is a, well, this is a policy decision, right?

Um, this is something that must be determined by the,

um, the, the business, right?

Mm-hmm.

the longer the RPO is the, the more.

Work you're going to have to redo.

So the question is, how possible is it that we can redo this data?

Right?

So if it's, um, I don't know if it's customer records.

If it's a, if it's a customer, uh, database of orders, is there some

other system that you have where you've, uh, you know, whenever you do

an order, you do a PDF of the order, you email that PDF to the customers.

How f how much effort is it going to take us to go back into all of our

outgoing emails from the the CRM system?

Look at all of the invoices for all the orders that we said we were going to

send, and then double check those against the orders that, uh, and you probably

don't have to double check it too hard.

You can say, the incident happened today at noon.

We had to recover to yesterday at midnight.

So we have all the, all the emails between those two different times

and, um, and then go and just reenter those orders manually.

That is, there is a cost associated with that, number one.

Number two, you may have systems where.

There isn't a backup, right?

You may have an e-commerce site that where customers can go to that site.

Put in requests and then that issues, um, you know, an an order

and now it's, no one's actually seen any of this stuff, right?

No one's looked at this stuff and then an outage happens that

that is irreplaceable data.

You're never gonna get that data back.

Right?

Um, and so again.

Th that will be measured both in terms of perception, uh, business perception,

Mm-hmm.

also there may be actual loss of revenue.

Now, quite possibly what will happen is you will, um.

Get a, a phone call from somebody going, Hey, man, where the hell's

my, where the hell's my thing?

I asked from my thing and, uh, it's not there.

Or like, uh, let's say, you know, I order from Amazon a lot.

I, I, I went back into my account and I know I ordered a butcher Majer

yesterday and it's not even, not only do I not have it yet, it's uh,

it's not even listed in my orders.

What the hell happened.

So you might get some of that business back, but it will, you'll suffer a.

A severe reputational, uh, damage.

Yeah.

Yeah.

And so this is the amount of data you could lose.

Now, I'm sure if you went to the business, right, and this is

coming from the business, right?

They're probably gonna tell you, I can't afford to have any data loss.

Yes.

How as a backup admin, are you supposed to respond to that question?

Well, you, you say, well, the first thing you say, well, our current ability.

Mm-hmm.

Right, based on testing, we've done testing, right?

'cause you're always gonna be doing testing, right?

Yes.

Okay.

yes.

So hopefully we've done testing and we've, we've figured out that

based on our current system, right?

So if you're, if you're having this, this conversation for the first time, right?

Uh, you know, I listened to this podcast and Curtis and PSA

said, I need an RPO and an RTO.

You ask it and they go, it's zero and zero, right?

You say, okay.

Good job.

Uh, thanks for giving me a number that I can work with.

And then you say, well, we can currently do three weeks, so.

Let's meet somewhere in the middle, right?

Um, and just like with our, with our to, we wanna see if we can pull them back.

But you, you should be able to pretty much, I mean, short of zero, right?

You will always lose some data and it will always take some amount

of time to do the restore, even if it's an instantaneous restore.

There's still some time, especially if we're talking a ransomware

attack, because again, you're gonna spend most of your time figuring

out what you need to restore.

You say to them, okay, if what you want is zero, which I'm gonna translate into

Or

less, less than, than one hour, right?

If you want less than one hour, RPO and RTO.

Then we're going to need to do this.

And this is, and, and I, I, I've gotten a ballpark number and

it's gonna be $20 million, right?

And then they go, okay.

Yep.

Right.

Or maybe they go, holy crap.

No.

Uh,

day is.

yeah.

Well, yeah, one, one day.

One day's good.

Either they adjust their expectations, right?

Or, uh, they give you the money.

Right.

Or, or somewhere in the middle.

Right.

They, they adjust their expectations, but they give you less money.

And you, you may be surprised, you know what they may do because it really,

the RPO and RTO are determined by how much money, what's the financial

impact to the organization going to be?

And you have to, like, if it's just a reputational impact, you

have to measure that in terms of.

Yeah.

Money.

And if you say, look, um, we're a company that currently generates

$50,000 a month in revenue, right?

So we're, you know, that's $600,000 a year.

Uh, we can't spend $3 million on a backup system, right?

Uh, but if we're a company that does $50,000 in an hour,

Yeah.

then uh, we can justify pretty much anything.

Yeah, the other thing to also remember is.

From a backup technology perspective, as you start to reduce your

RPO and RTO, it's not linear in terms of cost, It's exponential.

Like to go from like 24 hours to one hour, down to one minute, down to one second,

Yeah.

like it's a significant increase in cost.

Yeah, because you start, you start doing real time protection at that point, right?

Um, you start talking about things like, you know, continuous data protection

or near continuous data protection.

Um, or, you know, full, full, just full, um, replication without

really, because, you know, one of the things I often say is that like.

Replication's great.

And you could get a zero minute RPO or really close to it.

The problem is it doesn't go backwards, right?

So if you do need to go back even one minute, it's just

simply incapable of that.

So I'm not a fan of replication by itself as a, uh, as a protection

mechanism, but if you, if you have replication that somehow has the

ability to go back in time, which I would call continuous data protection.

Or I'd say a one hour RPO, it's so much easier to do than a,

than a one minute RPO, right?

percent agree.

Yeah.

Because with a one hour RPO, you take one, you take hourly snapshots, you

replicate 'em, you're good to go.

There are.

Myriad systems that will do that.

Right.

Starting from your former employer.

Uh, you know, you know, they probably, you know, NetApp, uh, probably

really perfected that I think.

Um, you know, and, uh, but they're, but they're now, you

know, a lot of fast followers that have that, that are doing that.

Right.

Um, and, um.

But if you want to, if the, the number of companies that do true

real time data protection down to the sub minute, that number is very

small and the price is very high.

Right?

Yeah.

Um, there's a lot of dead soldiers in that field, right.

Companies that tried to do it

yeah.

backed off, or ultimately got acquired for, you know, basically

it was like a furniture sale.

Yeah, my former employer happens to be one of those who's

very successful in that space

Nice, nice.

So the number one thing that determines your RPO is going to

be your backup frequency, right?

So if you are backing up once a day.

twice a day.

Mm-hmm.

Yep.

Yep.

Which, as you know.

Isn't always the case.

Is it always the case?

Well, and here's here.

Okay, here's an important question that I've always had.

Uhhuh?

Okay, so you finished a backup yesterday, right?

Say the backup.

it was a snapshot based backup.

It started at midnight yesterday,

Yeah.

right?

And it takes two hours to transfer the data.

Yeah,

at 2 00:22:25

00 AM your backup, your recovery point Objective.

at 2 00:22:29

Is midnight

at 2 00:22:30

yeah.

at 2 00:22:31

Right, because that's when

at 2 00:22:32

Well, your recovery point.

at 2 00:22:34

You're sorry,

at 2 00:22:34

Your recovery point is midnight.

at 2 00:22:36

Uh, you said it started at midnight and then it replicated it.

at 2 00:22:39

Yeah.

at 2 00:22:39

yeah.

at 2 00:22:40

Okay.

at 2 00:22:40

Now

at 2 00:22:41

Well that's assuming that because Are we taking it every hour?

at 2 00:22:46

No.

at 2 00:22:46

What today.

at 2 00:22:47

Oh, once a day.

at 2 00:22:48

Okay.

at 2 00:22:48

Yeah.

at 2 00:22:48

Yeah.

at 2 00:22:49

Okay.

at 2 00:22:49

So now the next backup will happen at the next midnight.

at 2 00:22:54

Right?

at 2 00:22:56

And until it shows up, which, let's just say it takes two hours.

at 2 00:23:01

The recovery point you use is a previous night spend night

at 2 00:23:04

Correct.

at 2 00:23:05

So technically, even though your backup frequency is set for 24 hours, RPO may

at 2 00:23:12

actually exceed your backup frequency.

at 2 00:23:15

Your RPA may exceed your backup frequency.

at 2 00:23:19

Yes.

at 2 00:23:19

Your RPA.

at 2 00:23:20

Yes.

at 2 00:23:20

Uh, so yes.

at 2 00:23:24

Right.

at 2 00:23:25

And because that's why I'm saying like the best you're gonna be able to do.

at 2 00:23:28

Right?

at 2 00:23:28

Uh, it really depends on when that.

at 2 00:23:30

When that actual incident happened.

at 2 00:23:32

So it's gonna be based on when the incident happened, it's gonna be based on

at 2 00:23:36

whether or not last night's backup worked.

at 2 00:23:38

Yep.

at 2 00:23:39

Um, do you do backups on the weekend?

at 2 00:23:40

I hope so.

at 2 00:23:41

Right.

at 2 00:23:42

Uh, because I, I've worked places where they, their last

at 2 00:23:47

backup was Thursday night.

at 2 00:23:50

Right.

at 2 00:23:51

And now it's Monday morning and they're gonna do their next backup Monday night.

at 2 00:23:56

Yeah.

at 2 00:23:57

If you have an outage on Monday and you did any work over the

at 2 00:24:00

weekend, you're gonna lose Friday, Saturday, Sunday, and Monday.

at 2 00:24:05

Right.

at 2 00:24:05

Um, so it, again, it's determined by your backup frequency and then, and

at 2 00:24:09

then any failures, uh, which again, only you, uh, can prevent forest fires.

at 2 00:24:15

Sorry, that's, that came out only you were gonna know what

at 2 00:24:18

your, what your actual, uh.

at 2 00:24:20

Yep.

at 2 00:24:21

You know, real recovery time or, uh, recovery success rate is right.

at 2 00:24:26

Um, and also, uh, you know, if, if backups get corrupted, uh, anything

at 2 00:24:32

like that, especially if backups get corrupted by, um, you know,

at 2 00:24:36

ransomware or anything like that.

at 2 00:24:38

Right.

at 2 00:24:38

Um, that's a, that's a good question.

at 2 00:24:41

So how do people actually test their RPO in order to determine their RPA

at 2 00:24:52

So, yeah, so good, good, good question.

at 2 00:24:55

Yeah, I, I, it's a little different than RTO, right?

at 2 00:25:00

And because really it's just, it's.

at 2 00:25:07

You, you don't really test it like, like you, because you, uh,

at 2 00:25:11

that's a, that's a great question.

at 2 00:25:13

You, it's, it's going to be the result of whatever your, your RTA is.

at 2 00:25:17

Right?

at 2 00:25:17

But you, you, again, it, it's more a discussion like how, how frequent are

at 2 00:25:24

our backups and how frequent are they?

at 2 00:25:27

Do they fail?

at 2 00:25:28

Yep.

at 2 00:25:29

Right.

at 2 00:25:29

Um, and then what you should do is you, you just like.

at 2 00:25:34

You, you report on what your compliance is, right?

at 2 00:25:38

Yeah.

at 2 00:25:38

And if, if the compliant, if the number starts creeping up or it

at 2 00:25:42

starts, like, hopefully you should, you should say, look, if a backup

at 2 00:25:47

fails more than once, then like all it should be all hands on deck, right?

at 2 00:25:51

Because it's bad enough that we're gonna lose, let's say, 24 hours worth of data.

at 2 00:25:55

Well now we're talking 48 hours.

at 2 00:25:57

And if it fails again, now we're talking 72 hours.

at 2 00:25:59

This is a huge amount of business data that you're losing.

at 2 00:26:03

So really.

at 2 00:26:04

It's not so much you can test, it's just, it's something you can monitor.

at 2 00:26:08

I think you can just monitor how well you're frequently backing

at 2 00:26:11

up and how well it's working.

at 2 00:26:13

and it looks like that's such a low bar compared to actually doing

at 2 00:26:17

like the recovery time testing

at 2 00:26:19

Yeah.

at 2 00:26:20

you should be able to do this

at 2 00:26:22

I.

at 2 00:26:22

easily.

at 2 00:26:23

Like there should be no excuse for you not to know what your RPA is.

at 2 00:26:26

Correct.

at 2 00:26:27

No excuse.

at 2 00:26:28

Um, and again, the better thing you can do to do RPA is to switch to,

at 2 00:26:32

you know, well, well, let's, we'll get to get to that in a second.

at 2 00:26:35

Um, yeah, absolutely.

at 2 00:26:36

Right.

at 2 00:26:37

So the, the first thing, again, this is like, uh, it's like

at 2 00:26:41

the 12 step process, right?

at 2 00:26:43

The first thing is to acknowledge that you're powerless over your RPO.

at 2 00:26:46

Okay.

at 2 00:26:47

Sorry.

at 2 00:26:48

So, so acknowledge you're an honest assessment.

at 2 00:26:51

Right of, um, of where you are.

at 2 00:26:55

Right.

at 2 00:26:55

You, you, you say you don't wanna lose an hour's worth of data.

at 2 00:26:57

We currently back up once a week.

at 2 00:26:59

Uh, this is a problem, right?

at 2 00:27:00

You have to do that.

at 2 00:27:01

Then you can rightsize the frequency.

at 2 00:27:04

You, you, you know, how quickly can you do that, right?

at 2 00:27:07

Maybe, maybe it's such, maybe it's, it's like, look, we currently

at 2 00:27:11

back up once a day, right?

at 2 00:27:14

Can we potentially back up, let's say.

at 2 00:27:17

Like, I don't know, during the day, right before the day.

at 2 00:27:20

Right after the day.

at 2 00:27:21

Um, you know, it depends on how your business works, right?

at 2 00:27:25

Um, could you potentially just tweak your, how frequently you can do it?

at 2 00:27:29

Um, and if you've got an incremental base backup system, remember that

at 2 00:27:35

if many cases, if not most cases, four backups throughout the day.

at 2 00:27:41

Take roughly the same amount of time as one backup once a day, right?

at 2 00:27:46

Unless what we're talking about is backing up the same data multiple

at 2 00:27:48

times because it's been, you know, changing throughout the day.

at 2 00:27:51

Right?

at 2 00:27:52

Databases.

at 2 00:27:52

Yeah.

at 2 00:27:53

Um, but like with databases, what you can do with databases is just

at 2 00:27:56

back up the transaction logs,

at 2 00:27:58

Yeah.

at 2 00:27:58

make sure that the transaction logs are getting backed up and sent to immutable

at 2 00:28:02

storage, uh, throughout the day.

at 2 00:28:04

That's the way you don't have to back up the whole database just

at 2 00:28:07

to get those transaction logs.

at 2 00:28:08

It may take longer to recover, but at least you won't lose the data.

at 2 00:28:11

Right.

at 2 00:28:12

Yeah.

at 2 00:28:12

Oh, that's a good idea.

at 2 00:28:13

Yeah.

at 2 00:28:14

Um, and then of course, again, backup validation.

at 2 00:28:17

Do the testing, see how long it takes, um, you know, you know, all

at 2 00:28:21

of those different technologies.

at 2 00:28:23

And then potentially consider, um, uh, a change in backup technology.

at 2 00:28:30

Right.

at 2 00:28:30

Again, either CDP or near CDP, uh, you know, the, the, um.

at 2 00:28:37

The, these are things that are your friend.

at 2 00:28:39

Generally speaking, many if not, most of those are storage based,

at 2 00:28:45

meaning that you will need to go to a different type of storage system in

at 2 00:28:50

order to get snapshot based back up.

at 2 00:28:52

That's not a hundred percent true, but there are systems like data core, right?

at 2 00:28:56

And I, I'm sure there are others where it can work with your existing

at 2 00:28:59

storage, but in most cases what people are doing is they're saying,

at 2 00:29:02

we're gonna buy Product X, right?

at 2 00:29:04

And, and we're gonna get snapshot based backup, we're gonna do

at 2 00:29:08

snapshots plus replication.

at 2 00:29:10

And just a just one final note on, on the RPO and sort of changes into technology.

at 2 00:29:17

Make sure you're taking into account.

at 2 00:29:19

Your SaaS applications, they're, they're, the RTO is gonna be very

at 2 00:29:24

different from SaaS apps, right?

at 2 00:29:25

Especially if the app itself is down, but you are, just make sure that you're also

at 2 00:29:31

looking at your SaaS apps like Microsoft 360 Fives and Salesforce where you're

at 2 00:29:35

generating data throughout the day.

at 2 00:29:37

Are there ways that you can incrementally back that up as well throughout the day?

at 2 00:29:41

The more modern backup technology that you're using, the easier it will

at 2 00:29:46

be to meet your RPO, uh, and because many, if not most modern backup

at 2 00:29:54

applications or SaaS backup applications.

at 2 00:29:58

They're doing deduplication based, replication based, very minimal

at 2 00:30:04

incremental backups throughout the day, stored in such a way that you

at 2 00:30:07

could very easily restore right up to the point of failure, assuming we're

at 2 00:30:11

not talking about ransomware, right?

at 2 00:30:13

Um, so just make sure you're taking all of the different parts of your

at 2 00:30:18

environment into, um, into play.

at 2 00:30:22

Any thoughts?

at 2 00:30:24

no, I think that's, yeah, I was actually wondering, 'cause in the RTO

at 2 00:30:28

episode, we didn't bring up SaaS app, so

at 2 00:30:31

Uh, yeah.

at 2 00:30:33

Well, because yeah, no, that's a good point.

at 2 00:30:35

Yeah.

at 2 00:30:36

I mean, again,

at 2 00:30:37

Yeah.

at 2 00:30:37

just whatever you have, whatever your environment is, you should be testing

at 2 00:30:42

recovery and of, of that thing.

at 2 00:30:44

Right.

at 2 00:30:46

Um, and, um,

at 2 00:30:48

not special.

at 2 00:30:49

what's that?

at 2 00:30:50

SaaS apps are not

at 2 00:30:51

They're not well, they are special and, and that people think they're

at 2 00:30:55

special, but they're not special.

at 2 00:30:58

They're just the same.

at 2 00:30:59

They have you, you are as responsible for that data.

at 2 00:31:03

And by the way, Microsoft finally gave in, they're now offer a

at 2 00:31:07

backup service at an extra cost.

at 2 00:31:10

To me, that's admitting the fact that.

at 2 00:31:13

You need a backup service.

at 2 00:31:15

Uh, and I would prefer, and again, nothing against Microsoft, right?

at 2 00:31:18

They, they do a great job with Microsoft 365.

at 2 00:31:21

I would still personally use a third party for the backup.

at 2 00:31:24

I would, and, and that's Salesforce.

at 2 00:31:26

Salesforce has a backup service.

at 2 00:31:27

I would use somebody else for the backup service.

at 2 00:31:30

Um, and that's not just because I used to work for one of the companies that

at 2 00:31:34

made, had one of those backup services.

at 2 00:31:36

It's just, I just, you know, sometimes.

at 2 00:31:40

Yeah.

at 2 00:31:40

When we read these stories about things that happen at vendors, we're like, oh my

at 2 00:31:44

God, I can't believe they did that thing.

at 2 00:31:47

Plus that thing, plus that thing.

at 2 00:31:49

And then of all of those things, it's like when, when I think about

at 2 00:31:52

like what happened at OVH in France.

at 2 00:31:54

And you're like, oh, it's that, that thing plus that thing.

at 2 00:31:57

Plus that thing.

at 2 00:31:58

So not only did they have like these container based storage things,

at 2 00:32:01

and not only did they have the, you know, and they were sharing power and

at 2 00:32:05

they were sharing, you know, and you know, the backup system was sitting.

at 2 00:32:08

Right.

at 2 00:32:09

You know, they said it was physically separate and by physically separate,

at 2 00:32:11

they meant it is over there.

at 2 00:32:13

Right.

at 2 00:32:13

It's, it's on the other side of the, the other side of the container.

at 2 00:32:17

You're just like, all of this logic.

at 2 00:32:19

When, when, when the logic is bad.

at 2 00:32:23

That bad logic can extend to, um, you know, um, and so again, not to pick on

at 2 00:32:30

Microsoft, but they're not perfect, right?

at 2 00:32:32

Um, this is a company when Microsoft 365 went down simply because somebody forgot

at 2 00:32:38

to renew the Cate certificate, right?

at 2 00:32:41

Uh, again, they're not perfect, right?

at 2 00:32:43

Um, so, uh, and the people that you have that administer the apps are not perfect.

at 2 00:32:48

So I, again, I would prefer to have it as a third party app, but.

at 2 00:32:51

Anyway, I digress.

at 2 00:32:53

All right.

at 2 00:32:54

Well thanks for chatting about RPO.

at 2 00:32:57

Thank you Curtis.

at 2 00:32:58

And hopefully everything turns out okay on the Tesla

at 2 00:33:01

Yeah.

at 2 00:33:02

uh, I might expect, uh, FaceTime, I'm guessing tomorrow maybe, maybe not

at 2 00:33:08

I'll, it'll be showing you the, the picture of a, the beautiful

at 2 00:33:12

underside of a completed Yeah.

at 2 00:33:13

Project.

at 2 00:33:15

All right.

at 2 00:33:16

Uh, thanks folks for listening.

at 2 00:33:18

Uh, I mean, if it wasn't for you, you know, I don't know why we do this.

at 2 00:33:21

So, uh, that is a wrap.